https://bauble.studio/ is a lisp-based procedural 3D art playground that I hacked together a while ago. It's fun to play with, but it's a very limiting tool: you can do a lot to compose signed distance functions, but there's no way to control the rendering or do anything "custom" that the tool doesn't explicitly allow.
So lately I've been working on a "v2" that exposes a full superset of GLSL, so you can write arbitrary shaders -- even foregoing SDFs altogether -- in a high-level lisp language. The core "default" raymarcher is still there, but you can choose to ignore it and implement, say, volumetric rendering, while still using the provided SDF combinators if you want.
The new implementation is much more general and flexible, and it now supports things like 2D extrusions, mesh export for 3D printing, user-defined procedural noise functions... anything you can do in Shadertoy, you can now do in Bauble. One upcoming feature that I'm very excited about is custom uniforms and embedding in other webpages -- so you can write a blog post with interactive 3D visualizations, for example.
Am I right that the output of the lisp code is ultimately a plain GLSL shader (like one might find on shadertoy.com)?
I built a SDF-based rendering system (2D) for my game, and one of the big hurdles was how to have them be data-driven, rather than needing a new shader for each scene or object.
Would be curious if/how you tackled that problem (:
Yep! It just outputs GLSL. It doesn't do anything smart -- it's a single giant shader that gets recompiled whenever you change anything, so it wouldn't really work for something like a game. I mean, it could handle like basic instancing of the form "union these N models, where N<256" but there's no way to change the scene graph dynamically.
I've done this for a project where the SDF functions are basically instructions, and you can build up instrictions on the CPU to send to the shader. and then the fragment shader runs them like a mini bytecode interpreter. You can tile up the screen to avoid having too many instructions per fragment. Kinda wild idea and performance may vary depend on what you're doing
The CPU builds an RPN expression (like "circle, square, union, triangle, subtract"), and the shader evaluates that in a loop.
I wasn't able to find examples of other people doing similar, but it seemed too useful to not be invented yet (:
Do you have any links to your work?
I'm writing some blog posts for my approach, but haven't finished them yet.
> You can tile up the screen to avoid having too many instructions per fragment.
I don't quite understand this part... If a given SDF needs N instructions to be evaluated, then how does tiling reduce N?
> performance may vary depend on what you're doing
Yeah, fill rate was not good enough with a straightforward approach, so I had to cache the evaluated distance values to a (float) texture atlas, then use those to render to screen. Luckily, standard bilinear filtering on distance values produces pretty decent results.
Yes sounds like the same thing! I also couldn't find anyone else doing it. Sounds super interesting what you're doing so I'd love to read your blog post when it's done if you want to drop me a message/email.
My project was using 2D SDFs for UI which meant you could use a bunch of primitive shapes and union/difference between them, and also add outlines, shadows, glows etc. This means that if you tile up the screen and use a union between two rectangles, only the tile with the overlap needs to calculate the union. It's a little more complicated in 3D with frustum culling.
I was doing it in webgl which doesn't have storage buffers and so I had to use uniforms to pass the data which is a huge limitation. Apparently webgpu could be better so I will try to figure that out one day. But it is early prototype so no links or anything yet.
I’m a tech nerd rare coin & currency dealer! I took my two hobbies and combined them into a real business and I’m having the time of my life. Just launched a proper retail site here:
About 50% of my days are spend doing the coin dealer stuff - hunting for inventory, buying collections/doing appraisals, going to coin shows and buying and selling in person, etc.
The other 50% I’m writing code and building out the tech stack for this business. I’ve written the whole backend for the retail site myself, which includes my own inventory management system, sync with eBay and other marketplaces, etc.
I’ve also built out a research tool which includes an ML price prediction engine engine (which sounds fancy but is really just a tabular regression model).
Backend is written in Crystal because I love the language and there’s nobody stopping me from using it :) Frontend is all Svelte and they’re glued together using a mini framework I wrote:
I probably have 5 years worth of ideas I still want to build and I wish I could spend even more time building it all, but it’s super fun actually using it in the real live marketplace so I’d never give that up.
Happy to chat about this stuff with anyone who’s interested or vaguely interested in numismatics.
I have a question - I've seen TikToks of people who buy rolls of coins from the bank and sort through them for rare imperfections then sell them on eBay. I've always wondered whether it would be possible to develop an automated system where a camera takes high res photos of the coins on a conveyor belt, compares to a DB of known imperfections and sets them aside?
Is anyone doing this? It's an interesting business model as the product is money so you'd only stand to make a profit never a loss.
Nobody that I know of is doing this, and see no reason why it wouldn't be possible from a technical standpoint. I think the only reason I can imagine NOT to do is that the ROI probably isn't that high in reality. Now, granted, I don't watch the coin TikToks because 95% of it is clickbait, exaggerate, etc. But my actual impression is that there simply isn't that much actually-valuable material out there hiding in bank rolls (despite what TikTok says).
Most of the people I know who do bank roll hunting and doing it because it's just kinda fun and there's a thrill when you find a silver quarter from 1964 (worth about $5) hiding in a roll of otherwise-normal quarters. But so much of the good stuff has already been plucked from circulation.
Having said that, nothing should stop a good hacker from doing something just for the hell of it :)
do you know why vendors take credit cards, square and applepay, even though those services charge several percent fees? part of it is for convenience for the customer, but another part is that shuttling cash around to the bank and back is time consuming, risky, and takes you away from running your business (let's say you are a breakfast place, you don't make your own cups and napkins or farm your own eggs and coffee either)
>product is money so you'd only stand to make a profit never a loss
you're grabbing the expense part of the business that everybody else is trying to shed. Let's talk also about time value of money. All the money that you've invested in cash is not making money passively as other investments do. Compared to putting the money in the stock market, you're losing 7% a year on this scheme, plus the expenses of running your business, and opportunity cost of not doing something else that generates income.
I did this for personal collections a while back and went through a lot of Canadian quarters to get one from each year and never even found the 1991 I was looking for which is somewhat rare. I guess if you do it full time or automate a bunch maybe you could make money, seems hard though.
One of my favorite design elements is the Guilloche patterning on currency, along with the history behind the use of Guilloche as a form of anti-counterfeiting.
Your site is very simply visually appealing.
Also - I like to order $2 bills from the bank. You can order then, mine delivers them on tuesdays - and they give you a stack of brand new $2 sequentially serialized bills. They are great for tips and gifts.
100% agree! I often get folks asking if their $2 are valuable, and tell them exactly what you do -- no, alas, they're not worth more than $2, but they are super fun to leave as tips because people still get a kick out of 'em.
Infra: I was hosted on Google Cloud for a while -- literally a single VM running Docker Compose, but I decided I wanted something a bit more flexible and interesting, so last month I switched everything over to Fly.io and I am incredibly happy with them. It's just so easy and fun to manage.
The retail site (rarity7.com) is just a small VM running a Crystal server process to handle web requests. Image hosting is all done on Cloudinary. My backend / inventory management / trading + research engine is a separate Crystal process in a separate VM. Both connect to a Fly Postgres DB. There's one other service which is a small python process on another VM which is doing inference on my regression model. That's super lightweight and I don't need any GPUs to do the inference (tabular data is nice like that).
Overall it's really nothing fancy and it works quite well. A few web-serving VMs and an inference service for my pricing model. I train/retrain the model a few times a year on a local box (my repurposed gaming rig running a 2080Ti).
I did it all in about 2-3 years. I was a big collector as a kid but didn't get back into it in a serious way until 2021. Becoming an authorized dealer is basically, fundamentally about building trust. The whole coin industry operates in a very old-school relationship-based trusted way. To become a member of most professional orgs or become an authorized dealer, you need to have 5-10 VERY solid references from other members or authorized dealers, and the only way to get those references is to be a part of the community, build trust, and build relationships by doing business with others (which means honoring your commitments and your word and writing good checks, etc.). In the end, you need to ask 5-10 people to personally vouch for you, so they'd better know you and actually trust you.
This is great, my brother is a history buff and a few years ago I bought him a Hadrian coin for his birthday. I'll look through this and see if I can find something else he'd be interested in, holidays are coming up and I'm always looking for nice gifts.
EDIT: is there any way I could set up an alert for when you add some non-US currency into stock?
That's awesome and a very thoughtful gift for him! Yes I'm planning on building alerting - BUT - I'm probably not going to expand beyond US-based stuff for the foreseeable future. It's just the niche I know and even though there's a big wide world out there, I haven't spun up my brain to learn about it all yet.
If there's any country in particular you're looking for, shoot me an email. I know a ton of other dealers in the business and I'd be happy to point you to someone who might have something he'd enjoy.
Awesome -- thanks! Tell your son to join the Instagram coin community if he hasn't done it already - there are tons of kids & adults on there and it's a real community that meets up regularly at coin shows, etc. Send me an email and I can give you a heads-up on some starter accounts to follow and get involved.
So let's say I have a few hundred silver dollars. Is this a tool to help me sort them, or does such a tool exist? Like take a picture and it identifies it, looks for common errors, and provides a base price estimate?
There really should be, but the fact is there isn't a tool that does this yet. It would not be deeply hard to build, but it would mean training a model which means getting good enough training data which means taking the time to actually do it. It hasn't been done because it requires someone who is deeply knowledgable about coins AND someone who knows how to train a model and build an app, and the fact is that intersection is pretty small (it's me and probably 20 other people? Give or take). This is very much on my long-term roadmap.
In the short term, the easiest way is to find someone who does this (eg me) and just email me some pictures. I'd be happy to tell you if you've got anything good. The old school neural net between my ears can assist.
That's fair. I have to imagine that there's a market for a product like that. My mind goes to auctioneers and estate sales. I know the local auctioneer would kill for a tool like that.
Something to keep in mind if you do build that: have a one time purchase or month long purchase option. When my wife's grandma died, we had to go through her coins looking for good ones. I would've killed for something like this but for a month.
Great question. Authenticity is easy - I mostly deal with certified coins, which means they’ve all been authenticated and guaranteed by a 3rd party service which stands behind their mark (they will pay you for the coin if they make a mistake, which does happen).
For shipping, that’s just pretty standard across most industries - I have shipping insurance and if USPS fails, they’ll pay for it. But losses happen and it’s just a matter of business (thankfully they’re rare).
Yes! Very much so. And in fact, the grading services are owned by the same parent companies. NGC and PCGS are the big coin graders and I think their card graders are CGC and PSA respectively?
I’m working on a collaborative ebook reading app. The idea is that you can create a reading group, invite people and then share comments and highlights and see each other’s reading progress.
It’s something I’ve been wanting for a while, for example to read a book with a group of friends or with a work team, but there’s lots of other possibilities including author reading parties, proofing and education. Got the basics of it working now, need to polish the UI and add the commenting and highlighting features.
I’m using Next.js and Supabase, neither of which I’ve used before so it’s been a fun but often frustrating process. Claude has been an amazing assistance, fixing my mistakes and countless type errors.
Ooooo.... Idea! One thing I've always wanted is an "asynchronous" book group. Basically, some way to tie the questions and conversation to a page or chapter, and then you can follow along at your own speed. Just passing the idea along since I'll never do anything with it.
Storygraph[1] does this with their Buddy Read feature. My wife and I use it to read books together and leave messages about different happenings, which only get unlocked when you mark that you've read up to that page in the book. It's a great feature, and we really enjoy it.
I wish there were discussion websites for media like this. If I've finally watched a tv show, everyone is already talking about another season. I want some sort of "season one insulated forum" or something. For all those that are in the same "temporal" spot.
That’s a cool idea, thank you for sharing. Yes, I want to tie conversations to specific locations in the text, but I love the idea of being able to set discussion questions as well. Will look into that.
Hey man, it is an amazing idea. I have been thinking to build something similar.
I am building reading clubs for my custom library - on top of crypto tech, and in the process, I have experimented with several book reader tech. pdf.js, muPDF and some other tools, which one did you settle on?
That’s a fascinating project, thanks for sharing. I’m using epub.js as it already has things like annotations and highlighting and it’s fairly easy to override the book styles.
From the research I’ve done, it seems that most ebook reader libraries are old and not very well supported. Haven’t considered PDF readers yet.
Is it going to be P2P? It would be much more easier to handle book versions and other such technocalities if it is, allowing people to share the EPUB via the platform and stay in sync
This is a hierarchical representation of any given piece of knowledge.
It starts with a tree root node that you specify (let's say Kung fu), then it branches out into multiples subcategories (techniques, styles, philosophy, weaponry, ect...) and then you can click on these subcategories to branch out even more into the graphical tree.
This is all generated on-the-fly with Claude 3.5. There is no limit to what knowledge you might explore.
The killer feature is that it is totally free and does not require to login.
Just click the link and have fun. I'll keep it free like that, as long as I can.
I hope you like it guys because it is the best project that I have up my sleeve.
As a marketing spin for this - consider packaging it into Nvidia NIM format and make it generate 3D graph as 3D OpenUSD scene. From where I'm standing this route has a lot of potential.
Also if you never looked into it there is a project called wikidata. There each object contains a unique ID and hierarchy, which helps build semantic web. Exploring their data using your interface might be effective. (please check similar projects though as an idea seems straightforward and someone might have already done that for them)
You're a legend mate I know nothing of what you say, but it is very interesting. It is time I brush up my skills and I thank you very much for the clues! I'll look it up.
Making a note to try this when it's not slashdotted (hackernewsed?).
This sort of thing is _very_ interesting to me, and I rather desperately want something like to this tied to listings of books.
I would like to read more diversely --- at one time I was trying to read one book from each LoC division, starting at the top/broadest (so A--Z), then iterating down, but this got to be exhausting because it was difficult to determine which book to read.
A later effort was to read biographies of famous people to my children in chronological order as they were growing up --- dry run was the very simple set of U.S. Presidents --- but again researching and ordering was a big problem.
So, my hopes for this are that it includes footnotes/references/links to supporting material (perhaps affiliate links on Amazon would be one way to fund things?) and that it is possible to click through to find information on specific technical topics as well as general explanations.
Thanks. I had tried that but the AI keeps inventing false books or providing fake URLs. When the AI will be proper I will include that feature, for sure!
A message to everyone: Thank you for your interest and comments and mails and votes and feedbacks. I am sorry the service was interrupted & down & I had to makeshift with ugly donation links in a ugly sidebar for hours, ect (I got 5€ donation! not crazy but it was actually useful)...
Anyway I managed to get it back running, and running strong! Mail me if you have nice words to say or if you think we can do partnerships. I received nice proposals already, which should improve the service significantly in the coming days. The goal is to entertain and educate, and as free as we can! haha, modestly. Thank you again for everything!
This seems really cool and I'm excited to play around with it once it's up and running properly again. These type of things are my favorite applications of LLMs.
A while back I made something similar in the form of an incremental "clicker" game where you split things ad infinitum: https://lantto.github.io/hypersplit/
It is back and running, and running strong! Enjoy. Your game is nifty maybe it'll give me ideas. There could be an option to hold Ctrl to split branches vertically without branching down, ect... cool
Nice work. An AI populated mind map. I like it.
Just missed a way to go from a node to more textual content if I want to know more than just a bullet list. E.g. an abstract-style paragraph, or even just a web search for the string.
Good idea mate I'll probably include a Amazon "search list" for books, and a video "search list" for Youtube, ect... Might be cool. If I try to provide direct links I get a lot of false results but I didn't think of adding web searches themselves.
(There was a feature to generate dissertations on whatever topic but I had to turn it off because I believe it was abused and that cost a lot of credits. I'll try to add it more securely very soon.)
Love this! Such an interesting UI choice. Would almost prefer new branches to appear further out to the right instead of overlapping existing words, and then auto-scroll me to the new spot to reduce clutter when browsing several tiers in.
The context sidebar is great! Maybe the first good "AI search engine" UX I've seen (most attempts copy Google look/feel which is more suited for web page results).
This is a very fun idea and I’ve had similar thoughts, without yet turning my mind to execution. Bravo. Is there somewhere I can follow more of your work in the future?
Hi cyrillite. Send me an email with "subbing" as object: pierre.treeofknowledge@gmail.com
Then I will gladly send you what I got, whenever I get it.
I will not forget. Thank you, friend!
I'll probably make a full blog soon for added convenience. (it'll be about AI products & production)
If you want a bite of some of my other work here is a AI game I made & which got totally ignored:
https://lywald.itch.io/the-fall-of-mankind
I tried to monetize it, that was a bit lame, but in the next few days I'll release it as free.
Well, since you said Kung fu, I started at martial arts and 5 clicks in it's telling me about the self defense applications of swinging from chandeliers. This is yet another pile of LLM word vomit.
> One innovative approach involves using ceiling fixtures to gain leverage and mobility. This concept can be particularly beneficial in scenarios where traditional movement options are restricted.
...
> Utilizing ceiling fixtures for movement in self-defense is an innovative method that capitalizes on environmental elements to enhance escape techniques. By integrating body mechanics with strategic use of surroundings, individuals can increase their chances of successfully evading threats in confined spaces. Always prioritize safety and practice these techniques to ensure readiness in critical situations.
I like this project but I don't think I agree on the 3 branches that I see when I open the website.
Science is not a standalone category. In fact science is directly connected to mathematics which is directly connected to philosophy (through logic for example)
We might argue that also art is an extension of philosophy (e.g. finding objective or subjective beauty).
Probably Claude doesn't agree with this definition.
I agree with you. Truth is this was the first branching generation which I generated, and so I kept it for the sake of history. If you manage to create a better "starting branching tree" which looks nice and feels nice then I might take you up on this.
I'm sorry guys but my API key is out of funds until tomorrow's or maybe tuesday's paycheck. I'm only a student and so my money is very limited. I'm sorry this is pathetic. I didn't expect to have so much trafic these last few days, and then with this thread. Come back soon!
What do you think that I implement BYOK (bring your own key), does that interest anyone so that it is never down for them but they have to pay Claude on the side? Is anyone interested?
Thanks, fastneutron! I might give you the code if you send me an email and you explain your intentions. No problem. For now, it is not full blown open source because I don't want the added headache of managing that (double checking security & quality... I have no time for this is a sideproject).
I have this creative writing website where writing friends can write branching fiction novels together. Sorta like choose-your-own-adventure, but more literary - third person past tense, actual characters and narrative arcs. Configurable number of choices at the bottom of every chapter, and as you're reading, if the choice label doesn't link to a chapter yet, you can write it. It's been floating around since 1996 or so, so my big project lately has been upgrading it. v1 was perl and gdmb files. v2 was php and codeigniter, and it's been limping along there for several years. When ChatGPT came out, I rewrote the whole thing for scala/play and that was super fun. Moved from ubuntu14 to ubuntu20, mysql5.5 to 8.0, etc. I'm done with the rewrite now but am still messing around with launching it, upgrading to ubuntu22 (done today), figuring out if I want to update jvm/scala/play before launching, etc. All this for only a group of six writers (so far)! But collectively we've written more than 425 chapters across a handful of stories, 450,000 words, and had a lot of laughs. I even wrote a snazzy graph visualization thing that animates the structure of the story maps. After I finish launching the thing on hardware/software that actually isn't EOL, I might open up membership to others.
Funny story, I had it released in the mid-90s. There are versions of it still on the wayback machine. You younger folks might not realize this but there was a time where there were a lot of fun internet happenings, and there was no spam. People hadn't figured out how to spam yet. There were also no T&C's or laws governing membership accounts. The first version of my site had no membership accounts. People would just create a chapter and assign a password for that chapter so no one else could edit it. And it was on GDBM, which could only handle one simultaneous connection.
Eventually, two things happened: the site got busy enough that people started experiencing errors when trying to submit/edit to the same story at the same time, and, spammers started realizing they could write chapters about certain erectile dysfunction medications. :) They also guessed some chapter passwords and overwrote them.
I didn't have a quick fix for this, plus I had ignored the site for a while, so by the time I realized, the site was overrun with spam and all my authors (who I had no way of contacting, remember: no user accounts) had disappeared.
So I took it offline for twenty years. I had copious backups, and over time I would reconstruct the story history. I had learned php by then from my career, and so I started rewriting the whole thing, which was honestly a huge project, especially because now every chapter has revision history. I had tried relaunching the site a couple of times without attracting any writers.
Then Covid happened and I was able to relaunch it and find writers. It's been fun since then. I have approximately zero commercial intent for this; I can't bear to turn on ads in the middle of an immersive reading/writing experience. Maybe someday paid accounts for extra features or a share of the copyright. My most fun idea for any sort of revenue is printed books for completed stories, or a mobile app "reader" with more features than ebook software. I think a wide release could be counterproductive since all the members currently kind of know each other, chat on Discord, and have writing nights. So it'll probably be a slow invite sort of thing.
I like this too! Kind of like those picture drawing games we used to play as kids, someone draws the head and neck and folds it over, someone else continues the lines... it always ends up in something crazy but entertaining.
I noticed my 2.5 year old niece likes zooming into photos of bugs on her mom's phone. I'm making her a little "game" where she can flip over rocks and find different bugs and other weird surprises.
A few months back, I posted an essay here about how I think we are entering an age where the personal library will become an asset class unto itself [1]. This is due to a combination of the advent of semantic search and the revival of personal knowledge management in the deluge of our age.
As such, I’ve been working on the software to bring this vision to life, which is called Your Commonbase (a portmanteau of Commonplace Book and Vector Database).
In short, the purpose of the work is to create a data structure that works the way humans store, retrieve, and share information. By making these three elements as close to zero stress as possible, you catalyze creativity through remixing and augmentation of memories. My hypothesis is a lifetime building a Commonbase creates an idiosyncratic system, filled with the interpretations of an individual or a group. This individualized structure then creates demand that others want. I.e, a curation of all of the books you have read, organized by the marginalia you have added to them. This is a system people would pay for, and also a system that becomes more valuable over time.
I’ve been “working in public” by posting updates on my site [2], and am just beginning a small waitlist alpha testing phase (email me if you want in!)
I've read a ton of your writing on this topic. There is still much more, but I'm tired. You described the problem well, and if your software alleviates it, then I see the value.
I'd like to try the software if it'd available please. But I'm not sure where it is or if it's out yet. Is it the Obsidian template you linked here, or is it a separate software called Your Commonbase? Because the Obsidian vault has a different name.
I have to say that the irony is hilarious that I can't find this information in the (well written) sea of information you posted. But then of course I'm not using your solution yet!
The problem with existing knowledge management systems is that they all eventually become victim to scale the operator can no longer manage.
Consider two scenarios:
Scenario 1:
You have a chaotic system of notes, dispersed throughout random pieces of paper and digital notes on your phone and computer. You have a relatively easy go of it when saving new notes (its as simple as pressing "New Note" or pulling out a new sheet of paper), but as you add more notes to your system it becomes harder and harder to find a particular piece of information. The cost of adding stays the same, but searching goes up and up. Scenario 1 causes us to eventually succumb to chaos.
Scenario 2:
You have an extremely rigid system of tag management, headings, sub-headings, sub-subheadings and sub-sub-subheadings. This taxonomy makes it easier to find information...at first. The problem with these systems is that they require a ton of manual maintenance, and they also make it easy to find certain "classes" of information, but fail at geo locating others. Much more perniciously, this scenario eventually stifles creativity as it falls prey to too much order. The cost of searching stays roughly the same, but the cost of insertion goes up and up.
(I am personally guilty of creating a system like this btw [1][2]).
Both of these systems eventually begin to become inert holders of information, as the processor (us), begin to fear them and stop working with them.
IMO, the closest technology to managing human information well in my opinion is well over two thousand years old, the commonplace book [3]. Simply put, a commonplace book is extremely resilient to chaos due to its centrality of information, but even people like John Locke had to create indexes to fully utilize them [4].
This changed recently with the advent of vector databases[5]. It turns out that commonplace book entries are the perfect form factor to benefit from an address in vector space, since entries are atomic. In simpler terms, the vector processing layer handles the order, allowing our system to "live" and assign headings/tags/etc. as it evolves. Vector databases love commonplace books as well, because many vector solutions have way too much noise as they chunk and store useless information at quite a disappointing ratio.
My system differs from current offerings because it makes no attempt to automize parts that are meant for the human, and makes no attempt at making humans do the work computers should be doing. Ergo, creating a type of symbiotic relationship.
Finally, a note on why I use the term "asset". An asset should become more valuable over time, and particularly, each individual component of the overall asset class should be worth more (e.g. $1 in a bank account of $20 is inherently less valuable than $1 in a $20MM bank account, because it grows slower). So in our scenarios above, the transmutation of information to knowledge peaks out at a logarithmic curve, subject to the scale issues I mentioned before. Old entries appear less frequently in even the most ordered systems, and when they do, it is only in one particular context. My system stores time of entry in the metadata, but since I use vector addresses, the information is accessible in many different ways (dog can be found when query == canine, fido, perro, mans best friend, etc...). An informational asset should scale linearly, and each action of create/read/update/delete should improve the health of the overall system.
There's much, much, much more I could say here, but I'll stop for now :)
I've been working on a kind of strategy game called Fall of an Empire (https://fallofanempiregame.com), it's a bit like a mix between Crusader Kings, Total War, Mount and Blade type games (just the overworld, not the battles), but rather than expanding you're trying to prevent an empire from collapsing.
It's a ton of work, especially with the number of systems - you've got combat, resource management, settlement development, food managment, espionage, diplomacy etc - these all need to play together well. And then you've got to add in the storyline, graphics, marketing -but I think I'm making pretty good progress.
I'm still using Unreal Engine 4 because I started work on that version and I haven't needed to upgrade to 5 since it's been released.
I've got a free prologue that I'm releasing on October 1, so now until then is a lot of polishing to make it work well.
Steam recently made some changes that are pushing game studios away from prologues. Instead, demos can have their own steam pages. Plus all hyperlinks will soon be banned.
I also noticed your game has been up on Steam since January 2022 and by the looks of it, you've done about zero marketing (steamdb.info). Hope that's on the horizon, it's a pretty darn important part of game development now!
Thank you! If you replace a prologue (as standalone game) with a demo, can you transfer the wishlists?
Marketing isn’t my strong point, if I ever had the money to make another the first person I’d hire would be marketing. My boss owns a PR agency and she said she can help in getting it out to journalists but that’s about it. I tried CPC and it came in at about $1/wishlist.
In case anyone has the same problem, I asked and they can't. Luckily I only have couple of hundred wishlists on the prologue so I'll switch it out easy.
This looks fantastic! Love the premise & well done for what looks like an incredible solo effort. I was tech lead on Ozymandias & have another minified 4X in the works at the mo & putting out feelers with publishers etc. Sounds like you plan to self publish this one but I’d love to chat & see if there’s any way I can help with your launch, whether just through advice or via introductions etc. You can reach me at biggles (at) bravobravo.games though I may end up finding a way to contact you directly first :-)
In all seriousness, there’s a constant nagging in TW community for more historical games. Pharaoh didn’t cut it for the wast majority of players. You could capitalize on that, the genre would benefit from that.
If this one does alright my next would be a crusades game with similar-ish mechanics, feudalism, the pop system from Victoria 2, multiple-settlement provinces and realtime movement.
They recently released Pharaoh: Dynasties which is pretty much a complete overhaul of the base Pharaoh game, and is pretty much what the game should have been.
They gave everyone half their money back, this free upgrade, new factions, bigger map and a ton of changes. Overall its been a big push from their side to keep the player base happy.
But yeah, there's obviously a big market for historical games, from city builders, combat, etc. and I'm here for them all.
A safe generalization about Dynasties is that “no one cares”. It has about 2.5k players now, same as Attila (one of the most controversial titles released 10 years ago).
People keep bringing up Dynasties, but it’s obvious CA doesn’t know how to make successful historical titles anymore. They dropped the ball even on 3k which had some kind of potential.
I started in uni during the start of Covid because I was playing Warband and I wanted to see if I could do something like the overworld in Unreal. Then I left it for a few months, got the idea of turning it into a proper strategy game and worked on an off until this year when I really started working on it. I probably put in 2-3 hours on weekdays after work and 6-7 on weekends. I think that I'm definitely more productive now because I have more experience, and I had literally no money at all at the start due to not having a job.
This looks great! It seems to solve the issue I always had with civilisation where the game becomes progressively easier as you advance ahead of the AI, leading to a late game which involves lots of management but little risk.
I wondered this myself and found some. I've played: When Rivers Were Trails, Thunderbird Strike, Dialect, Zulu Dawn, This Land Is My Land. There are a lot of options in boardgames (maybe more than with video games?), such as Burn the Fort.
Most paradox games enable this to some degree. It's pretty limited in terms of imagining what this could look like outside of the european conceptualization of civilization progress but it's there.
A feedback: the art and assets feel like they are AI generated but an interesting game idea, not sure if it makes sense as most the need to defend an empire that took me days to build greater than a procedurally generated one.
Right now I'm mostly thinking about AI. Right this minute I'm sitting in a cafe reading a book on Multi-Agent Oriented Programming with a framework called JaCaMo[1], and later tonight when I get home I'll probably spend some time getting my Fuseki[2] server loaded with some base schemas (SKOS, FOAF, etc) as I slowly start working on getting things set up to explore some ideas around integrating symbolic logic with LLM's.
And I took my new quadcopter drone out and did some flying last night for the first time. As in, my first time flying a drone, ever. The results were... predictable. Let's just say, I bought a cheap (< $100) drone for a reason. This thing will wind up destroyed. In less than an hour I managed to crash it into fences, walls, bushes, cars, dumpsters, the ground, an armadillo, Elvis Presley, a 1974 AMC Gremlin, and Nickelback. Well, more or less.
It brought to mind this famous scene[3] from the movie Days of Thunder:
Harry: I want you to go back out on that track and hit the pace car.
Cole: Hit the pace car?
Harry: Hit the pace car!
Cole: What for?
Harry: Because you hit every other god-damned thing out there and I want you to be perfect.
I’m working on a screen play called DRONE RACER about kids who race human sized quadcopters through the subways of NYC. It’s a mix of Hackers and Fast & The Furious.
Using a lot of off the shelf AI and pushing the limits of what’s possible.
If HN supported images in replies, I'd totally be posting that "shut up and take my money" meme right now. As it is, I'll just say "that sounds awesome, and I hope it sees the light of day eventually."
As an experienced drone pilot - if you can, DJI is unbelievable at stopping you from crashing. Start there to get the feel of it (and more importantly not get frustrated and give up)
If you want to learn more freestyle without assistance, i can't recommend enough getting a simulator on your computer and crashing a digital drone a million times first. It's much cheaper. I think there is a lot out there now but i used Liftoff for many hours before i tried to fly a proper drone
I was wondering. I have a DJI and it is stupid simple to fly. I’ve yet to crash mine but I don’t fly it a ton. I’m certainly not trying to fly it through complex paths like an open doorway and into a house or something.
Yeah, I'm flying a super basic drone with none of the built-in crash avoidance / navigation / etc. stuff for now. I'll upgrade to something fancier later, but for now I'm totally satisfied with something I can crash repeatedly and when it breaks, throw it away (or salvage parts from it to build another drone).
I'll look into the simulator thing. I didn't even realize that drone simulators were a "thing" so I had not gone looking for anything like that.
I would highly recommend getting a drone simulator off Steam. You can practice the controls and drill them into your fingertips during times when you can't fly the real thing (battery recharging, night time, weather, ...). Most advanced radios allow you to hook them up to the computer so you can fly with the exact same inputs, but a console controller is also acceptable.
Making fingerboards. Little skateboards you can do tricks with. I'm always fingerboarding at work when I take a break from programming.
It's a ton of fun, especially writing Fusion 360 scripts to do all the parametric modeling of decks and molds. Then 3D printing molds with different parameters, pressing decks with veneer, making art for the decks, packaging, etc. It's an incredibly niche hobby but I've always found fingerboarding and making fingerboards to be infinitely creative.
A friend of mine started with fingerboard more than 20 years ago, and also selling half pipe for fingerboarding made by wood. He got a degree on architectures, created a company to make half pipes and he made de skate park for Paris Olympics, xgames, redbull etc.If you are curious, take a look on @insaneramps on Instagram!
Yes! You can make obstacles pretty easily, too. I made a rail with a drawer pull I got at Home Depot screwed together with two blocks of wood. I use it on my desk all the time.
The cool thing is, you can fingerboard on almost anything. Get a stack of books and you can make a stair case. Get a piece of a granite counter top sample and you can make a ledge. You can use a bathroom sink as a skate bowl. There are also lots of companies making fun obstacles with different materials. The possibilities are endless.
I'm working on a web app that helps you discover books, movies, TV shows, video games, and songs that you'll like. The app makes it super easy to describe what you're looking for and then gives you a unique set of 10 suggestions on each run.
It's the first solo project I've done in many years and I'm having a great time learning. I've also been able to get some great recommendations for myself and it's fun to use!
However, I gave it "Jane Eyre" (a mid-19th century novel by Charlotte Bronte; it's a classic) with the expectation that it will return "Wuthering Heights" – another a classic novel written in mid-19th century by Emily Bronte (yes the authors are related) – among other things. People who like Jane Eyre tend to like Wuthering Heights, and vice versa.
But everything it returned were new works – published in the last 20 years – of stories that are set in the 19th century.
The only filters applied were:
"Preferred: Gothic romance novel, Victorian literature, Highly influential classic literature, Early 19th-century England ".
So, something seems amiss for now. Nevertheless, it's a very cool idea.
(Edited to fix my grammar; English is not my first language, so pardon me for any mistakes.)
This is happening because I intentionally added a strong bias for more recent books. I thought that's what most people would be looking for - at least for fiction - and I also assumed there were other good ways to discover older books (lists, collections etc).
You're absolutely right though, it should give you some Victorian lit given your preferences! I'll work on addressing this.
This is amazing, bookmarked and will try and use it to discover some new artists.
It seems to reverse the song/artist names occasionally. For example I've got a recommendation of 'Russian Circles' by Harper Lewis, but it should be 'Harper Lewis' by Russian Circles.
It is also telling me the song 'Ten Billion People' is by As I Watch You From Afar (who have a song called '7 Billion People all Alive at Once'), however it's from Explosions in the Sky.
How to find "Military science fiction series" that are "Complex and Strategic" and includes "Space battles and tactics", "Action-packed sequences", "Strategic dialogue", "Multiple book series expansions", "Futuristic space environment", "Spaceships and battle fleets"? Use yogurrt. Never found it so easy to narrow down my interests and find new books. Thanks!
Bookmarked. It actually does seem to work well for the items I have selected. I actually got some interesting hits I feel compelled to look up now ( nice call on adding purchase link -- unobtrusive way to monetize it ).
This works surprisingly well. Just shared it with my sci fi group.
Idea: make an email opt-in and then new releases based on old preferences. My use case is always to go to DVDs/netflix releases, filter for sci fi and then high IMDB.
The results of the categories look similar to what happens when prompting ChatGPT with "describe a list of {genres/moods/etc.} for <content>". Also the links are all for query pages on amazon/spotify/etc, which says LLM to me (rather than a database). This dynamic schema generation -> UI supported query builder seems like a really interesting direction. I'd love to see it for the web more generally and I bet Exa.ai+LLM of choice would be a pretty good place to start. If it's an LLM I'm curious about hallucination rates and excited for future web-level RAG in only returning real objects.
Also you can supply content which doesn't exist and it will automatically infer genre and such (another strong indication of LLM). I.e. you can imagine the aesthetics of the kind of content you might want by title and then use the UI to explore real recs related to that.
PS to the creator: I love the simplicity of the design and the execution is inspiring. Sent a dollar for my searches :-)
You are spot on :) - including about potential additional and broader use cases for this kind of UI. Hoping I can explore that more. Thanks for the tip about Exa.ai!
Also, I'm sincerely flattered that you like the design and execution. Thank you for that and for your contribution!
This is a huge gap we have these days. Everything out there is incentivized to suggest the most selling stuff rather than unique and interesting stuff.
Yeah, this is definitely an important aspect of the problem I'm hoping to solve. I've found that selecting interesting preferences yields some interesting, non-obvious results "out of the box" - but if you really want to go off the beaten path try selecting the "Little known" preference under Other (and make it Required).
It's going to have online text courses with interactive examples and coding exercises, but I'm also in the process of adding video tutorials. These videos will be of 2 types: ones where I teach you the theory, and ones where we actually build a project from scratch.
I feel like CSS has always been something that was made to look harder than it actually is. In its essence, the syntax is very simple, and the vocabulary is quite basic. There are only a few things you need to know to be able to code an attractive and flexible responsive web page. For comparison, I find programming backends much more difficult.
Even though I've been working on this project for almost a year, I decided that next month will be the day I finally launch it.
Sounds great, I know a lot of people who could benefit from this.
I think CSS has historically been a lot harder than it was supposed to be, because browser support was shit. They were slow in adopting new CSS features, and existing features (even simple features!) weren't correctly implemented in all browsers (looking at you, Internet Explorer!).
Things are better now, for the most part, but everyone seems to have migrated to web frameworks, and writing HTML and CSS from scratch has become something of a dying skill.
Sounds nice indeed! By the way: Ironically, I’ve found a few CSS issues on the landing page (input placeholder is cut off on mobile viewports) and your personal website (the blinking line is overlapping the opened menu on mobile). Hope it helps!
I'm working on FreeInChicago, a site for finding all the free activities to do in the city.
I love Chicago and the incredible wealth of free festivals, museums, public art, music, parks, workshops, and so many more activities available to appreciate. I've found there isn't a good way to discover them though so I'm building the tool that I've always wanted. Right now I have a landing page up, have done some work on sourcing data, and I'm working towards defining a discovery interface.
Nice! I'm a fan of SF Funcheap https://sf.funcheap.com/ and always miss it when traveling to other cities. Another up-to-date local resource is something to look forward to.
Thanks for sharing! Seeing someone building something similar and being successful is incredibly energizing. I had doubts that it would be something folks would actually be interested in using.
Nifty! I worked very briefly about 12 years ago on a similar concept that was live at chicago.com. Things didn’t work out with it for various reasons but while it was up I found it quite useful. Some of the original team is still in Chigaco and building stuff. Best of luck shipping it!
I am building an Excel extension that lets me use LLMs in formulas. For example, I can write =PROMPT(A1:E1, "Extract keywords") in a cell to extract keywords from a row and drag the cell to apply the prompt to many other rows. I find this useful when I want to use AI for repetitive tasks that would normally require copy-pasting data in and out of a chat window many times.
I actually built it for my girlfriend who was writing a systematic review paper. She had to compare 7.500 papers against inclusion and exclusion criterias. She obviously did this manually because she cares about scientific integrity, but it sparked the idea to make an AI tool to automate repetitive tasks for people like her who would rather avoid programming. Now I just find it useful myself for a lot of ad-hoc analysis tasks like prompt engineering, rag tuning, and comparing model outputs from anthropic, openai, and google.
Yes I am really interested in good use-cases, send me an e-mail at kaspermarstal@gmail.com! It’s called Cellm, implemented with ExcelDna and is fully functional for Anthropic models. Currently preparing the repo for public release and writing some docs.
I work in a scientific field, and I was analysing a large amount of data related to clinical trials. Part of the work required categorising the data fields, and while traditional Excel approaches were helpful, ultimately I had grind through a large number of fields, manually sorting/categorising and Googling details for about half of the entries.
I wondered whether an LLM could be helpful for some of the steps I was taking manually (hence a quick exploration for such a tool) but there didn't seem to be an easy way to test this. (I could have pulled it into Python and tried to automate it that way, but I didn't time/energy on that occasion.)
Thanks for sharing! If you want to try it out, it will be published to github.com/kaspermarstal/cellm in the coming weeks. Will reply to this thread when it is live.
I recently created a custom 220gsm pre-washed cotton fabric by twisting thrice a 60s yarn (60/3 Ne) into 20s before knitting. It could never be commercially viable so released something similar.
Essentially combining the softness of fine count cotton with the weight and durability of a heavier garment.
I've bought a lot of different T-shirts and I'm always interested in trying to find a better (more durable; more fashionable; more comfortable) one. My current favorites are the Buck Mason Field Spec t-shirts https://www.buckmason.com/collections/mens-tees — could you explain how your shirts might compare? I'm willing to buy one just to try it out, but since you're the owner, figure I might as well ask you for more information — especially why you created a custom cotton fabric.
One thing I like a lot about Buck Mason is that their T-shirts are made in America from American-grown cotton, are any of your T-shirts made in America?
EDIT: just ordered a variety of your shirts, looking forward to trying them out. Very easy checkout experience!
Buck Mason makes great tees. Our flagship tees use 100% American-grown Supima Cotton and are similar in weight to their classic Pima ones.
The Heavy Crew is closer to their Field Tee— ours is likely not as heavy but softer and more durable due to twisting two fine cotton yarns into one before knitting. They’re made in our own facility in Southern India.
I’ve been refining our fabrics for nearly ten years to get the best fit and feel, sticking with cotton for its comfort. More on our fabrics here: https://www.marchtee.com/us/difference
Thanks for commenting on the checkout! We’re gradually rolling out stateside, so any feedback is welcome—just reply to the order email anytime.
Great work, for some unknown reason I cannot bear wearing non-cotton shirt and cloth, but cotton based clothing is far from durable. The best cotton shirt that I ever had until now (in terms of wearability and durability) is a mecerized golf cotton shirt by PGA Tour (casual wear not for golfing) but now I cannot find any golf shirt (PT included) with the same material anymore. As far as I'm concerned you have got a winner there mate if your cotton shirt is as good if not better than the PT shirt, hopefully your clothing materials will be available globally off-the-shelf.
Do you mind explaining your sentence here?
>It could never be commercially viable so released something similar.
We do mercerize some of our products, though it’s more costly with regulated heat. If you're near an M&S, they carry a mercerized Polo, but I find it too crispy.
> It could never be commercially viable, so released something similar.
Apparel is about scale. You can make a great tee, but fewer people will pay $100 versus $30, so finding a balance is necessary. Cotton can age well, but the trade-offs (fiber blends, uneven yarn, spirality resistance, no ribbed neck) often show in tees under $50.
The recent relicensing of Redis to a non-open-source license bothered many in the community. But the groundwork for the relicensing was laid much earlier. I've been working on relicensing monitor to track various projects attributes that can affect the ease of relicensing a project.
I also think CLAs are eerie and goes against the open source spirit, I don't think CLA alone puts a project in "high risk". I'm not sure about the FAANG open source projects that are used as libraries (Guava, React, ...). These projects fundamentally don't jeopardize these company's businesses, and serve to increase the developer goodwill amongst the engineers. Nobody can predict the future but I can't imagine these projects becoming relicensed.
More plausible scenario is them becoming an abandonware, but even in those cases the community can carry the torch.
Agreed, and there's a few projects with CLAs that are ranked lower due to mitigating factors, like K8s [1]. I honestly don't get why they have a CLA, anyone know?
The impact of developer good will is difficult to measure, so I don't attempt it. Redis burned community good will so badly with their relicensing that several forks rapidly emerged. Seemed like a predictably poor decision to me.
I also don't want to pick favorite companies, because it's subjective, companies can change strategies or even sell projects off. What if Meta decided to sell React to a patent-troll-like company instead of just abandoning it?
> I honestly don't get why they have a CLA, anyone know?
There are valid reasons to have a CLA: confirmation that your contributions are not encumbered by an employer contract is a good one. What there is rarely an excuse for is a copyright assignment, which often gets bundled into a CLA.
The only non-nefarious example of copyright assignment that I can think of is the FSF, but only because they have such a strong record on software freedom.
do you consider the likelyhood that a project will be forked?
for example react is listed as high risk, but at the same time i would consider the risk to be very low in the sense that if it is relicensed then it will be forked immediately backed by a community that is strong enough to sustain such a fork.
i'd go as far as saying that for react relicensing would kill the project because the majority of users would go with the fork.
for other projects the risk is higher because they don't have a strong FOSS userbase that could sustain a fork, and because most current users would not care.
What happens after the relicensing isn't measured as the community has already been disrupted. I'm also less certain on how to fairly measure and predict fork likelihood.
One challenge for forks is that relicensing doesn't always jump from fully-open to fully-closed. There's a lot of "fake open-source" and source-available licenses, like the one Redis now uses. These may "only impact you if you are AWS", so a fork "that AWS can use for free" feels less compelling. If React was to relicense, I expect they'd similarly take a small step.
Is there a specific reason that Firefox is considered low-risk for a rug pull? In my view, Mozilla doesn't seem the same as it once was but maybe there are specific reasons the open-sourceness isn't in jeopardy.
Firefox scores well because it uses a copyleft license and the ability to relicense contributions remains with the original authors. Mozilla can't unilaterally relicense the Firefox code base as they haven't been granted that ability by the contributors. The copyleft license means they can't slap a new license on top (like a permissive license allows).
The rating criteria was designed to consider legal facators, like license terms and CLA, so concerns like Mozilla buying an ad company aren't factored in. Those concerns feel more subjective to me, but are certainly valid.
I imagine Firefox would die instantly if it moved to a restrictive license. It would be too easy to simply switch to a popular fork, and Firefox's userbase are the type that would be compelled to follow through with it. Even ignoring FOSS principles reasons, most folks are browser-savvy enough to understand the implications of such news -- their favorite browser is about to kill itself so they have to pick another one.
Also consider that Firefox would no longer be the default browser in most Linux distros, and likely prohibited from official package repositories entirely.
A new CI/CD platform. Local CLI (run without git push), remote debugger, automatic content-based caching, DAG-based definition, and dynamic tasks. It’s taken a lot of work to build it, but I’m really excited about how well it’s working. https://rwx.com/mint
I haven't tried this yet, but I wanted to say I think there's a lot of potential in this space. There's so much friction with the current popular solutions...and yet it's so hard to justify trying some of the newer and less popular ones.
I wish you luck because there are a lot of good ideas in here. Running locally and remote debugger are the most exciting to me.
I like the idea, but I have used Azure Pipelines, and the key differentiating it from GitLab.ci is its "service connection" that enables seamless integration with third-party services and a wide range of tasks (PowerShell, pscore, az, docker, etc).
I wish your dependency semantics yaml could be converted to Azure pipeline task, job, and stage similar to how typescript converts to javascript.
I am saying this because there are very complicated infrastructural problems that hundreds of people work on at big companies when creating a ci/cd I see you are not focusing on them yet but that DAG logic could be applied to existing azure pipeline yaml.
This seems really neat! I'm a huge proponent of incremental CI for any project that gets big enough to care about test speed.
I'd be pretty worried about an abstraction ceiling on the caching here relative to something like Bazel, though. Like would Mint get mad if I generated 4000 dynamic tasks?
Out of interest: CircleCI also has flaky test detection, rerun failed tests etc. Or did you mean that the CI that you were moving towards could benefit from Captains features?
I didn't make myself clear I think. You mention that you love Captain, and wish you had access to it when migrating away from CircleCI. I'm interested in understanding the specific reason(s) for your love for Captain's features, that seem to also be available in CircleCI?
I haven't been working on much at all lately, struggling from bore-out at work. 8 hours of boredom a day is surprisingly exhausting, and it's spilling over into the rest of my life. Starting a new gig next month, hopefully things will pick up again.
I had a bit of a tough time towards the end of my last job, and felt a similar ennui. If you can, I’d highly highly recommend picking up a completely unrelated hobby to fill your non work (and occasionally work, when you’re in that limbo) time. I couldn’t believe the effect it had on me, and it meant that I started my new role feeling refreshed and optimistic, rather than just diving into it like I normally do.
Personally, I hit the gym for 4 months, mostly to fix the lack of sleep that came with the restlessness and lack of mental exercise.
I’m working on a new game. I’ve been trying to get back to the basics of the adventure genre. A fun and engaging story, choices, strategic combat and a simple, straightforward approach.
No microtransactions, no advertisements. No networking! The game is meant to function completely off line. The game will start with a mobile release in probably October and should be available on Steam shortly after.
I’ve been using Flame, Flutter’s game engine to develop my app. My artists are amazing and have been providing fully animated sprites using Spine. I’ve used Flutter before but this is my first time getting into Flame.
If you’re curious how it’s going, hop on the mailing list at https://danger.world to learn more. I’ll be looking for my initial playtesters here in the next few weeks.
I'm writing a book. A novel. By hand (this is, no AI involved, just me and vim).
Sometimes I feel I'm loosing the time, because nowadays there is a lot of AI generated content and even more competence in self-published books.
After a long walk against myself, of about 10 months, it's nearly finished (in my native language, Spanish). It still needs a few more reviews and retouching.
I got recently unemployed after +20 years as Linux sysadmin, and my wife is now unemployed too (after +20 years in HHRR), fortunately we have still a few savings.
I dream that it will (economically) work, but most of the time I intuit there will be less than a few sales from family and friends.
Depending on how it goes, I've already the script for the second and third parts.
In parallel I'm researching different ways to generate cash flow without working for another person. I would like to avoid going to search for a job in the current market of cloud, docker and kubernetes, as I'm more a hardware/colocation guy, and 99,9% of job offers request for docker/kubernetes.
Mainly I use vim. But yes, sometimes I also use a traditional notepad with a soft ink pen. To avoid myself some screen time, because I'm out of home, or just to make some use of it.
I think the output flow is about the same using pen and paper than using vim, at least for me.
The original "By hand" did mean "without AI help/inspiration".
I'd love to read when it's available or in preview mode! The mental organization I think that develops when writing on a notebook is far worth the trouble of typing it on vim later, especially for fleshing out scenes and redrafting.
Not sure if call it mental or physical organization, but there is more crossing out of words and whole sentences, words that I try to fit into small spaces, or arrows to/from outlined areas below with more space. Like:
A technique that I don't know what to call, is the horizontal brace. Like the ASCII 123, but horizontally, to insert something between two words, above or below the line.
A handicap of the notebook, is that you can't find historical data/names/dates so easy, or synonyms... but as I iterate a few times over the text after the first draft, those gaps can be fixed latter while typing the text.
I am working on improving a system to inform citizens about the public benefits for which they are eligible [0]. It's hold by Barcelona City Council.
It analyzes the social situation of a person, or group of people that live together, and forecasts those benefits that they could access. It is an old base code that uses OpenFisca for the back, and React for the front. However, it's quite fun though. Currently I'm working on make it more stable as there are still some parts that could break the process.
Hey! It's the English version, something got broken. Your message made me learn about that. Thanks! Spanish and Catalan versions seem to be working properly.
On my next deploy that will be already fixed. Unfortunately, software done for the city council has very slow cycles due to several actors interacting. And the deploy will delayed specially now, on summer. Today I'll fix that, but probably you won't see changes until next week.
I’ve been working on Don’t Double Book Me (https://dontdoublebookme.com) the past few weeks out of frustration with keeping my personal and work Google Calendars in sync.
Previously used Reclaim but found $10 a month to keep 2 calendars in sync was excessive and the software was increasingly becoming more team oriented, no longer for individuals. Felt like I was paying for a product and also the product. I just needed to keep 2 calendars in sync, not smart meetings, analytics, integrations, etc. ideally I set it up and forget about it.
I really needed a way to sync my calendars privately, without all the extras. Now that Dropbox has purchased Reclaim, it's even more important I feel like my calendars are not spied on. I knew I could provide a similar service.
Here’s what makes me excited about it as a user:
Affordable: Just $20/year with a 7-day free trial. Cancel at any time for a pro-rated refund.
Privacy First: No need to store events on a servers enabling unlimited calendar syncs.
Working Hours: Adaptive feature that adjusts if an event falls outside your specified working hours.
Round Event Times: Opt to round event start and end times to the nearest 15 or 30 minutes for cleaner scheduling.
Invitations: Block time between calendars when you receive a meeting invitation that you haven't responded to yet. Decline it? The time block on your other calendar is removed.
My plan is to keep it small and focused, while also listening to user feedback for how you'd like to manage your personal and work calendars more efficiently.
I don’t have a personal calendar and my work calendar ends at 5. I think your market is a lot smaller than you think. The kind of people who ask friends to book time on their calendar are considered freaks anywhere outside of tech hubs
Don't Double Book Me isn't some crazy startup idea that needs funding, I am not shooting for the moon with this. Feature development was basically done in a few weekends and I like that it solves a single problem well. There is definitely a market for people who need a tool like this. Look at Reclaim, Calendly, Clockwise, CalendarBridge, OneCal, etc, which have successfully tapped into this market but are focused on selling to teams.
I designed it for individuals balancing personal and professional commitments. I work somewhere with a heavy meeting culture, nothing I can really do about that. I found myself double booked often with personal commitments I had. Consider a scenario where you need to attend your child’s school play, but you forgot to also block off time on your work calendar, now your colleagues think you’re free and invite you to something. You've been double booked.
Events outside your working hours aren’t synced, protecting your personal time and privacy. The default setting is to create events on a destination calendar with the summary explicitly being "Busy" without any other information, but you can change it so it mirrors the event details should you choose.
I put a generic “out of office” block on my calendar if I have personal business and it’s rare enough that I don’t get why it needs a solution. Maybe other people are scheduling things during work hours a lot more often than me?
That totally works if it is rare for you! The last 2 months I've had a few things I've added to my personal calendar that synced with work calendar: lunch with a family member, dropping off or picking up car from service, haircuts, contractor stopping by, dentist, vet appointment, etc. It's nice to not worry about it since it marks me as Busy right when I create the event.
This is a nice idea, but assumes both your employer and your personal account use Google Calendar. If that assumption goes away I might be interested in this.
What guarantees does this make about access to data and how does it make them?
I've mentioned my project a few times around HN, but might as well reiterate :)
I took some time off from work to teach myself Rust and to build a WASM colony simulation game. You've got a colony of ants, they're in a cold, foggy crater, and you help them grow and survive. The simulation runs 24/7, like a Tamagotchi, but a bit more complex, like a simplified RimWorld or Dwarf Fortress. I am hoping to design game play systems which focus on mental health, self-care, addiction, motivation, and personal growth and to use the gameplay as a means of encouraging awareness in the player.
I haven't added any features in a while, unfortunately, but it's on my mind. For a while I was just adding whatever popped into my head, as a means of learning Rust, and I naively thought the full idea would crystallize with time, but it hasn't. So, I'm trying to take a step back and figure out how to actually make a coherent game that does justice to the mechanics I want to see in my simulation. I've spent a lot of time thinking rather than coding, but I'm optimistic that I'll get through this phase sooner or later. I will admit, though, that trying to take a novel approach with game design is overwhelming at times. That's okay, though! I'm enjoying the process of tinkering with the project and will likely continue tinkering on it for many years to come as a creative outlet for self-actualization.
> The ants are on an alien world, the user is in a satellite orbiting the planet, and the two establish a symbiotic relationship, out of necessity, as a means of terraforming the planet.
That reminds me a lot of Helliconia books trilogy. Congrats for it! I'll look for some time and give it a try.
Tip for those looking for some more details on the mechanics of the self-care aspects of the game: Some details are present in the google doc linked in the repo README.
Hi! This year I quit my corporate job to work full time on https://crucialexams.com/. This month I am working on expanding our new offerings at https://pmpready.com/ and https://vitalnursingexams.com/. Our platform provides a fun way for students to test their readiness for industry certification exams like the PMP, BCEN, CompTIA, AWS and more. Written in ASP.NET, MySQL as backend and good old Javascript/Bootstrap for the frontend. Nothing fancy, works great!
Officially retired. Managing the money I saved to retire on, which is more time-consuming and stressful than expected. Working on yet another static site generator but stymied by my terrible Git/GitHub skills. Studying historical Western secular music. Jazz guitar too.
Happy site-maker-makin'... It is a most respectable yak shave. I had such a good time making mine a couple of years ago in Bash (https://github.com/adityaathalye/shite).
If Git/Github are interfering, just drop the folder in dropbox (or suchlike) and version control like it's 1990 (ssg_2, ssg_3, ... ssg_final_final :). Nobody can fire the officially retired developer for doing this in 2024.
Making swim baits. I started carving and paints fishing lures during Covid. It’s easy to knock out something that looks like a lure, but it’s surprisingly nuanced to make something that behaves in the water as you want it to. Mass production is absolutely the correct way to go about this, but - you know - it keeps my hands busy in the evening and I’m improving my carving, airbrushing, weighting/buoyancy, etc with each one.
I have a ton of respect for people who can craft beautiful and functional baits. I watch Nate Marling a lot on Youtube and it amazes me what he can do with a block of wood and some wire.
Do you fish your lures, or is making them just a hobby unto itself? If the former, what species are you targeting? I consider myself a bass angler, but I honestly haven't a. had much time to fish lately or b. had much luck catching bass when I do find time to fish.
I’m in Minnesota, so mainly targeting panfish, bass, and pike. Salmon when I can get away for the run. I do fish them, but I’m only now getting to the point that my crank/swim/glide baits act right. Hand carving right now, but may try to start making molds next.
Engineered Angler Lures is another YouTube creator worth checking out.
Also, I try not to be precious about them. Give them away to friends. Don’t stress if they break off. Just make another one.
I’ve been building a personal finance app that runs fully in the browser (using the automerge crdt and sqlite) for over a year now at https://tender.run.
Recently I’ve been taking more of being able to flexibly run sql against this data, and this past week I’ve been working with d3 to make fancy sankey graphs to show income/expense flows. Quick preview here: https://demo.tender.run/reports/sankey
That looks quite nice and the local first approach was a good idea.
I wonder how it will compare to ActualBudget and its pre/post rules and the GoCardless account data importer.
Side note: The first col of all text overlaps with some other text on mobile devices (both landscape and portrait mode) and thus is unreadable.
Also the inbox didn't work an I can't look at the inbox as the filters stay open.
Edit just noticed that this is a SaaS. If it would be a single payment for a license or sponsored FOSS, I believe it could work. The current way, I would assume that it will meet the fate of all the other apps in this non-market. Good luck though!
Pretty interesting take on a finance app! I like the inbox and filters on transactions. I'm not sure I understand the benefit of storing it all in the browser though. Does the data just live in the browser? Does anything store on a server?
Originally this was a privacy angle - the data is primarily stored on your device, with backend storage that’s treated like backup and sync only. I had plans for e2ee that built on this but it didn’t turn out to be a big differentiating factor.
Working in local-first turns out to be really nice for making the app feel super snappy. The responsiveness you see in the demo is the performance you can expect in day to day usage.
I recently bought a used Seadoo and discovered the dirty little secret of personal watercraft that they don't want you to know. Unlike cars with their standardized ODB2 ports, those aren't a requirement for PWCs. And the diagnostic scanner hardware/software costs hundreds or thousands of dollars. Major affront to the right to repair, so I'm ending up doing some CANbus hacking.
That’s kinda surprising. In Polaris ATV’s they seem to have this built into the interface of the machine itself so you can read the codes on its LCD. I don’t know how complex they are but I assumed they were a bit more open than OBD-II instead of less.
I've never seen an LCD screen in an ATV; my neighbor repairs small engines and stuff like riding lawnmowers and ATVs - i hang out from time to time and he's said more times than i can count "i got no idea what's wrong with it, probably a wiring issue somewhere i gotta track down" or somesuch. Some of the machines he's repairing cost upwards of $40,000 so i'm not sure it's a price thing.
My point is, i'm sure he would have mentioned how great or how awful such a debug console was, at least once, in the decade i've known him.
> For more information on diagnostic codes and display code definitions for your vehicle, contact your Polaris Dealer.
uh huh. MIL, FMI, SPN? anyhow the next time i'm at his house i'll look at any of the polaris vehicles there and see if any of them actually have useful diagnostic information. canbus and odb-II are standardized; i expect the polaris codes to be "ohhhhhhhh. you should bring that in to the dealership straight away" regardless of what it is.
You can tap a few buttons and get trouble codes. At least on the 2018 or so Sportsman I had. This feature was described in the manual, if I remember correctly, and all the trouble codes were well documented. It would show things like “P0012” which then said “high voltage to xyz”. It was very similar to OBD-II systems I’ve used.
I'm trying up standup comedy this summer -- open mics for now.
I thought it would be absolutely terrifying to be in front of a bunch of strangers and try to make them laugh, but it turns out if you're prepared, it's not that hard. Open mic crowds are benevolent and don't expect you to be the next Richard Pryor or George Carlin anyway, and I don't "engage" the public at all; I just tell my jokes.
I try to come up with new material each time so there's some work to do, but it's fun.
If it doesn’t psych you out too much, trying to engage with the crowd maybe once per set or every other set could be a good way to force the habit and make it easier over time.
Not that it’s necessary for all styles, but when I’ve been to comedy shows in the past the people who are able to deal with the crowd really well are always the most impressive to me :)
Good luck! I am happy to know that you don’t do crowd work, which has become fashionable now. I remember Big Jay Oakerson’s roast where someone pointed out that crowd work is what you do when you don’t have any material :)
Crowd work almost always comes down to race or profession of the audience. It can be funny only in so many ways.
My first job was in a FinTech and the way production access was managed scared me. This is my approach at streamlining the process.
Basically a PR review flow for SQL queries, enforcing the 4-eyes principle so you never accidentally can do a Delete * form users, forgetting the where clause.
Thank you! This was initially a startup idea but it was very hard to sell a cyber security solution with a very bare bones product to companies large enough to have this problem. Because the required features of such large orgs are just too much for such an early stage team.
So I'm now doing it a bit as a side project but I hope to at some point also offer a paid version to fund development long term.
Yes, I already thought about this. In theory all "production access", be it ssh, k8s commands, database access of any kind or even a ruby shell could go through such a a flow.
It's a bit of a pain to integrate well with everything though, but I have a lot of plans, will just take some time to mature :D
Working on adding better phrase matching for Marginalia Search.
The way position used to be stored in the search engine was approximate, using something like a bloom filter. With this change, they'll be stored in an exact fashion using a gamma coded positions list instead.
In the phase of dotting t:s and dashing i:s at this point. Hope to have it in production in September.
Compared to the majority of what I read below, I'm a rank amateur and have no right to post.
I'll do it anyway, as I need to learn to own where I'm at and have a backbone. :)
I'm learning the Zola static site generator and using to to build my blog at https://jeff-mitchell.dev. The focus is my mis-adventures learning the Rust language.
I've found it a challenge to get off my feet with Zola, but I'm slowly figuring things out. Little victory this evening, finally figured out how to get images linked in posts to render properly.
A graphic novel about a future world run by benevolent AIs who have found that the best way to get humans to behave in easily-predictable ways is to manifest as extremely unctuous clowns.
Thanks! I really need to get around to making my Wordpress installation email about new stuff.
Right now the best way's either to follow it on Patreon (no money needed) or to follow me on Mastodon (@anthracite@dragon.style). Or fire up your RSS reader and point it at the site.
The insight is that if you read text from a tiny enough box right below the front camera on the MacBook, it appears as if you're talking to the camera.
Boom! Easy Eye Contact With Camera.
I scrunched up the Notion app and placed it as well as I could in the safe area to test it out for a couple of videos, but then I just wrote a web app because it seemed like the next logical step.
Once you paste the script and hit play... you can only see the text in the safe area so your eyes don't wander.
Real teleprompters have a person feeding the text at the perfect rate, for pauses, and so on. It'd be an interesting experiment to feed the audio to Whisper or such, roughly match it to the text shown, and control speed based on that.
Oh, wonderfully simple, might use it for recording voiceovers. Might be useful to add some kind of syntax for pauses, like a blank line between paragraphs.
I've been working on a site that helps you find in-person work in NYC that is actually convenient: https://walkablework.com
I recently left a startup I had cofounded after a few years because we were a remote team and I came to the conclusion that we weren't going to be successful without working together in person. I wanted to make this site to help people like me use location as an early filter to find good companies and teams to work with. Let me know if you have any feedback!
There's a site called SwissDevJobs.ch that similarly maps tech jobs down to a neighborhood level. You might want to compare notes, especially on company transparency & community calendars.
I think this is pretty incredible, and lots of potential. I simply love anything that's displayed with a map visual representation. Do you have a team you're working with on this? Shall we chat privately? ping me here - sibynyc@icloud.com
Hey guys! We're engineers/designers from France, and we've built the Ultimate DIY Battery that you can repair and refill!
It works with 90% of the bikes/motor brands on the market, so I assumed that some people here might be interested, if they got a non-functional batteries but they still want to use their e-bike?
We believe that everybody should have control about stuff they own, and we should fight against planned obsolescence!
Here are a few videos about our founder on the battery itself, why we built it, and how to assemble it:
We'd love some feedback from the e-bike DIY builder community
Oh, and it's launching as a Kickstarter in September and there is an offer for early-backers here https://get.gouach.com/1 for a 25% discount on the battery!
I thought the reason the packs are sealed is to stop amateurs from swapping in cells of differing voltages, which short each other out and cause sparking/fire hazards. There's a reason li-ion battery charging is current-limited, they have no internal resistance.
As I understand it, it's somewhat the opposite for internal resistance. You need to have current limiting in place because the internal resistance of the cell is turning that current into heat, and heat is the enemy of batteries.
(Probably there's also a complicated chemical reason for current limiting that I don't fully grasp)
This is why our Infinite Battery comes with a casing that we designed to resist the heat of a fire, thanks to it's thick aluminum casing, and a design which dissipates heat quickly.
And YES THIS WORKS! We have put a massive effort in making our casing sustain the battery fire, and extensively tested it (videos coming soon), so you can sleep safely (and us too!!). See our complete fireproof testing on our documentation/FAQ
Prevention is better than cure though. plus you've got to consider cell gassing/venting/explosion risk/toxic fumes as well as fire prevention.
If they were doing per-cell protection, fair enough (cheap 18650's don't have cell protection usually), but I think they are still connecting 4 cells in parallel, with only serial-bank MOSFET's like in usual sealed packs.
> a casing that we designed to resist the heat of a fire, thanks to it's thick aluminum casing, and a design which dissipates heat quickly
Thus turning a difficult to deal with Li-Ion inferno into a slightly less difficult to deal with block of superheated aluminum, that will happily set on fire anything that looks at it funny?
This is always my concern about fire-proof metal casings for batteries: where do you put that so it doesn't set anything on fire?
Also spot welded connections work well, transfer reasonable amounts of current and don't tend to vibrate off during use. We tear down packs at my job and there are warnings inside some not to re-use cells for vapes, they really don't want you taking these apart. Larger packs like EVs and E-bikes are probably worth refurbishing and repairing.
Indeed, we recommend swapping all the cells at once! But using our battery it means that you recover cells in perfect conditions that you can use in second life applications!
Super cool! Any plan on making different form factors? For electrified cargo bikes, the battery is better placed bellow the cargo [0] so a squarish and flatter pack would be nicer.
This is a great idea. I'm actually researching DIY solutions to electrifying my boat (I was a USAF electrician, once upon a time) and the prepackaged battery solutions are so expensive and inflexible. (Granted, for marine use there are some extra considerations...)
There is a guy that rides the trail near me thats riding an ebike with a triangular battery pack that looks identical to one some HNer built here - He's too fast and I havent been able to stop him.
Another question for the DIY e-bik crowd, is what kind of OTS motor could one harvest, like a motor from an electric lawm mower, or edger, weed-wacker? Are there power tools where the electric motor is useable? (like how people originally used mower engines in goKarts and such?
Also, for fireproofing - mix Powdered Aerogel into a Silicon Slurry and dip the battery in that - the silicon and aerogel will make a heat sleeve thats quite fire resistant.
Thanks for your message! The battery is already designed and in production, it's running on 1500 free-floating e-bikes in France without issues since around 2 years
We're doing a Kickstarter because we're a startup so we need the upfront money to buy the parts and build the battery! We expect to ship to Kickstarter customers in November 2024!
Since you have the ESP32 - can you emulate the Apple Airtag on it please.
I have a $7,000 Orbea E-bike and I have an Airtag hidden in it. Its a great feature.
Not sure if I could redo the Orbead Battery this way because its internal to the frame - but it does have an extended battery pack option - perhaps this could be wired to use that connector as your interface allows for any ebike it seems?
The post recently on HN about the flipper zero being able to emulate the airtag is pretty nifty, and you could use these guys effort:
The Goauche BMS Tesrt Bench needs its own video/HN post - and this is exactly what would be great info to get via a LORA messaging for all the telemetries from the pack?
I'm the CEO of Tile and we are working on a line of products for theft.
As a genuine question, I'm very surprised that many people don't seem concerned that AirTags alert thieves to their presence. If your bike is stolen the thief will get the notification, even if they are on Android (with the new shared platform rogue tag detection).
I'm trying to understand the user psychology here - not pitch our own products.
Do you have any specific features you would like to prevent theft?
Great to meet you. I have a ton of thoughts on this matter, but not well threaded for quick reply: so forgive the stream-of-thought nature of the following.
There is no real secret answer to stopping bike thieves, because the primary problem is both cultural and infrastructure.
We dont have bike culture that allows for safe bike leaving-about, and we dont have bike infrastructure for safe bike locking-about. (Bike Hack SF: Take your nice bike into a nice hotel and tip the DoorMan $5 to check your bike into luggage check rather than lock it up)
So, my airtag is in a case with security bolts under the bottle cage - and its more a check that I can know where it was last seen. Though I didnt remove the chirper.
What if you could have a TILE that is the fob for this battery. Where the battery wont go unless the paired Tile Fob is present?
A Lora Tile would be killer If you had a Lora network on this battery, with a TILE app infra on top of it and all the batteries talked to eachother - you built an anti-theft-mesh so all the bikes tiles all tell eachother about themselves.
Tile integration with Shimano Motors and e-tube iso app to stop the motor from running unless Tile fob/phone app present. A magnetic Solonoid crank-shaft-blocker.
And above all else pushing for better bike policies in municipalities.
We are doing a satellite integration with Hubble Network - 24/7 global coverage, no increase in battery life. It sounds like scifi but 2 satellites are up. Give us a year...
I want to build things using Instant DB... I keep finding ShinyObjects here, but - Tiles using a LORA Network which updates a csv/json style payload and transfers GPS through lora messages.
I think there a lot of interesting things to make bike mesh a part of a security fabric - but ensuring that there is no way to spy on location data thats not yours? Doubt thats possible...
The fact that something is being tracked and the thieves know it’s being tracked is a deterrent itself. Like the value of stolen iPhones crashed because activation lock made them worthless. So they moved on to Samsung Galaxies, until they implemented their own activation lock.
If you steal a bike with a tracker on it you’ll probably leave it right there for the real owner to find.
Of course I know this, but I am trying to go deeper. With AirTags and the new standards you literally can find and disable the tag for anyone using the Google or Apple networks.
I am surprised that people think the benefits of AirTags outweigh the downsides given the alternatives, of which we are one. I'm mainly just trying to learn.
I also think you probably are overestimating the deterrent value. Thieves are not necessarily thoughtful and the alerts aren't real time. Once they get the alert they might ditch the bike, but if it is back at their garage they will probably disable or remove it.
I also am a bit triggered by this line "Honestly, you as the CEO should know this." which is on the edge of being an ad hominem. Why did you choose to include this? And wouldn't you know that I know these things? Like I am asking questions to get nuanced user feedback. Do you think someone who runs a consumer product company (I started Life360, Tile's owner), isn't deeply aware of how customers think?
First, I want to apologize. The last statement came out a lot more combatant than I realized. So I removed it.
I was a Tile wallet user until I stopped carrying a wallet completely. There are 2 types of thieves: the professional and the opportunistic. Opportunistic, the majority, aren't doing more than they have to and if the bike has a tracker they know about it makes it less appealing. And if you can get the casual thief to give up immediately then the device has already payed for itself.
The professional will chop it immediately. I see kids in my neighborhood with $2000+ Trek Marlins, with all the expensive mountain bike parts removed. They simply wanted a bike to freestyle on. And doing all that probably took hours if not days. The professional is not going to spend anymore time than they have to disable a tracker when there are better options all around.
Im curious how Tile addresses this concern for the safety of individuals themselves in the opposite direction. I know people who have been stalked and assaulted before. Does tile have protections against this for people hiding tiles in your vehicle or bag?
It's a very nice concept and I think it's much better than the held-together-with-duct-tape concepts usually seen in diy bike conversions.
That being said, I am a bit skeptical about your claims to be fireproof. Did you try to start a battery fire inside your enclosure? Aluminum melts at about 600C, while lithium burns much hotter than that.
Helping others with their mental health (after my own struggles).
Worked as a software dev/manager for a decade, went through workaholism, burnout, then alcoholism, depression, all that. Doing a ton better now, and taking some time off to write about what I went through and hopefully help out others going through the same thing some: https://depthsofrepair.com/
An Open Source Hybrid AI Search Engine: Instantly get accurate answers from the internet, bookmarks, notes, and documents. Obtain the most precise answers in the shortest time. With one click, AI indexes your personal knowledge base, eliminating the need to remember or manage it.
I'm working on an invite-only, personalized medical education platform called MedAngle, particularly for emerging economies. I started it in my third year of medical school, and we're now at 90,000 doctors/medical students, along with over 100 million questions solved and billions of minutes spent studying and excelling.
I get to lead a team of 175 doctors and students across premed, medical, and dental education. I am the first doctor + full stack technologist in the country. It's super rewarding. No funding, just off our immensely low price point that things are still growing quickly. All software written in house.
I and a friend are working on Elixir Emporium [0], a single-player crafting RPG that uses layers of generative AI to create and inject new game data as you play.
It started out as just a prototype asking what happens if you task an LLM with generating crafting recipes for every combination of items in a game (which was already super fun), but it's exploded into all kinds of crafting, harvesting, and item manipulation systems that literally weren't possible in games just 5-10 years ago.
Now we're working on NPC simulations based on last year's Generative Agents: Interactive Simulacra of Human Behavior [1] paper. Dialogue and persistent memory are obvious, but we're also using the ReAct pattern [2] to give NPCs an influenceable decision loop that dictates what actions they take throughout each game day. And there's some other fun stuff like quest generation and using LLMs as a decision engine to determine if certain player actions complete these dynamic quests.
There's still a lot of work to do to make it feel more like a polished game, but we've been focused on the underlying systems and getting them feeling great and I'm really excited to see the game come to life.
I'm working on something similar but instead of basing my efforts on the generator agents paper I'm basing it on a technique I call narrative generation. It requires far less context and fewer prompts, and focuses on generating narrative aspects and letting the traditional game engine things simulate the remainder of the world.
As an example, with the system I am building you only need to input what action the player has taken, and all the NPC's actions and dialogues will be generated.
Secondary applications for med school - once your initial single app gets processed, every school (out of the 20+ you apply to) wants multiple essays from you. The process of preparing for and applying to med school is a nightmare and consumed the last two years of my life. I've had a lot of really cool life experiences because of it though, learned so much about medicine, and done some actual good. But I just want to get in this cycle and be done with endless essay writing.
I've been working on and off on building an app for learning chest x-ray interpretation, but that's shelved until I finish applications.
I've been working on an iOS app called snapmail: lets you take any photo on your phone, filter / sticker it, and send it as a 4x6 postcard anywhere in the U.S. for $2.99. Fun way to stay in touch with friends & fam. One of a few no-account-needed, single use iOS apps I've been working on.
I've been building a web extension for an Inuit language (ᐃᓄᒃᑎᑐᑦ ᖃᓂᐅᔮᖅᐸᐃᑦ).
Most Inuit in Canada speak Inuktitut, which is a language with long words and two different writing systems: the latin alphabet, and syllabics. Syllabics are specially adapted for Inuktitut and well-loved by the Inuit, but unfortunately can be a pain to input on a computer, so often times the more cumbersome latin alphabet is used in casual writing.
Tom Scott has an awesome video about how syllabics work[1], but breifly, the shape of the character determined its inital consonant spund, and the rotation determines its vowel sound. So ᐱ = pi, ᐳ = pu, ᐸ = pa, ᑎ = ti, ᑐ = tu, ᑕ = ta, etc. The word "Inuktitut" becomes ᐃᓄᒃᑎᑐᑦ in syllabics.
Transliterating between the two is fairly simple, but there are edge-cases around dialects and whatnot. The more interesting problem from a technical perspective is having a web extension that can detect Inuktitut on a web page (in wither writing system), and transliterate that into whatever writing system the user desires, whilst never accidentally transliterating the other text on the page ("inhabitants" could be picked up and transliterated as "ᐃᓐᕼᐊᖯᐃᑕᓐᑦᔅ", for example, even though that makes no sense).
The project is mostly using Rust via WebAssembly, which has been a lot of fun to work with and has let me do some awesome things, like avoiding heap allocation and using compile-time hashmaps to do conversions on the text. The build system has to do a lot and I eventually settled on python. Right now I'm trying to wrangle JS and the DOM (there's a lot of edge cases to deal with), and that's been difficult as it's not my wheelhouse.
Could your project also extend beyond DOM, to Canvas and WebGL code? I could imagine some very interesting creative uses in a game engine setting. Also: check out Aya by Cohere, a Canadian AI company.
I think you may have replied to the wrong comment or misunderstood my project, it's mostly finding text from a certain language based on some (fairly simple) heuristics, then converting it from one script to another in a systematic way, with some options a user can set. I doesn't have much relevance to LLMs or game engines, unless your game has an Inuktitut translation, in which case it'd be better to transliterate ahead of time ;)
Inuktitut is very cool. I'm not a speaker so I'm relying heavily on research, but I've always loved making tools for syllabics and working with text generally.
Then congrats all the more on this work, I had assumed a motivation from attachment also. As for the tech stacks and langs yeah they're fun, but I meant in how abstract Inuktitut looks, I guess the human language in the loop is the real meat of the problem and of the "working with".
If you're curious, I don't have any direct connection with the Inuit but I got into syllabics because I have some Ojibwe speakers in my family, and Ojibwe also uses syllabics. None of the current speakers I know can write in Ojibwe (in any writing system) due to their residential schooling, but we have letters from the generations that have passed on.
I made a few syllabics converters for Ojibwe when I was younger, but Ojibwe and Cree are harder to build converters from than Inuktitut because they have a lenis-fortis distinction in the latin script but not syllabics, so the word "anishinaabe" can be converted unambiguously to ᐊᓂᔑᓈᐯᒃ, but trying to convert the other way could render any one of anishinaabe, ani*zh*inaabe, anishinaa*pe*, or ani*zh*inaa*pe*. I don't think it could be done accurately without some sort of machine learning. There's also a lot more complexities surrounding how different dialects are encoded. In the end, Inuktitut was simpler to work with and has a lot more everyday speakers and text to work with, especially because the Canadian territory of Nunavut publishes its government documents and websites in Inuktitut too.
There's a lot to cover but I think the technical challenges might be interesting to the Hacker News crowd, I may make a series of videos about it someday!
Yes it is right up my hacking alley. I'm working too with two-way asymmetric conversions between languages/charsets as noted in this thread. It's interesting to have such a niche tech challenge associated with your ethnic group of languages. In Manitoba are any of those spoken?
Ojibwe and Cree are both spoken in Manitoba, but it's too far south for Inuktitut.
I read through your other comment on this post and read about the work you're doing. I didn't quite understand –I think it's hard to explain a lot of these linguistics tools briefly without uploading a lot of context– but it seemed up my alley as well.
> It's interesting to have such a niche tech challenge associated with your ethnic group of languages.
That's true, I think it can be a lot of fun and rewarding for technologists. Kevin King is a font designer who was tasked with making a new font for syllabics, then found that some syllabics had never been adopted into Unicode and set about getting them approved[1]. He worked with the communities and the Unicode committee and got them into Unicode, and that must have been really rewarding.
There's difficulty in approaching from the outside, though. I'm trying to be careful to provide value and not overstep. The Inuit especially can be defensive, because you see a lot of people online who sort of fetishize the culture, or people (certainly linguists) whose work can seem extractive to the tight-knit Northern communities.
I've always wanted to build tools working with text that people use on a day-to-day, but I also worry that this will be just another project I spend months on that will not see any use. I'm trying to make connections so that I can build value without seeming extractive. It's hard to know where to start, however.
I wouldn't consider your months spent on it in any way wasted, even if you have native critics, it's a kind of ad-hoc anthropological resource that you add to the pool of tools available in the web, and on the long run I think it will have a good impact.
Thanks for your review also and yes it's still early days, tons to do but of the fun type. I'll look into Manitoba as I have relatives which may be related to the cultures you mentioned, thank you!
When you say syllabics it reminds me of syllabaries (?) of mesopotamian languages and hieroglyphs, but I guess it's just the terms sounding alike.
I'm working on android app that exposes camera stack trough http/websocket API.
The idea is that phones have very nice cameras and lot of media processing power and such API could enable hackers to easily incorporate capable camera stack in to their projects without android development experience in any language they like.
There are millions of working phones in the drawers battery backed, with plenty of computing power, nice cameras, hardware acceleration, good connectivity (WiFi, ethernet over usb), that could get new purpose.
Previously I was working on automated photogrammetry rig and this is perfect use case for that.
If you see other use case for such API share your ideas.
If you think that's cool idea to develop and would like to help write me at hn@zelo.pl
I checked droidcam before and AFAIK it solves the problem of using adnroid device as webcam using server and client apps.
I want to expose low level camera capabilities like manual focus controll, capturing raw images, iso, etc. DroidCam exposes only subset of functionality I need.
In short i want it to be more like programmable controlled digital camera than webcam.
BTW scrcpy offers similar features trough adb to DroidCam.
I'm working on a plasmid (vector) editor called PlasCAD, for molecular biology. I'm doing it mainly as a learning project, and trying to fill in the gaps where I found limitations in existing software. As of last week, it's in a state where the main features are there. I'm going to now focus on polish, and adding specialty features. Would appreciate any and all feedback!
It lets you view and edit plasmid sequences, features, and primers, and has some tools like automatic primer creation for cloning, primer QC, protein sequence viewing, and interop with common file formats.
I’ve been working on robotics for a few months now. I have built an arm, started receiving parts for my quadruped, and have learned enough RL to at least hold opinions on it (my background is ML/LLM training, so a bit different).
Eventual goal before I return to work next year is to have a robot I can take on walks with me that will pick up trash.
When the hype was focused on chatbots, I joined a startup that built one focused on ecommerce. It turned into a useful product that actually help real customers.
Looking back, there wasn't any material to actually build a useful chat agent that resolved real-world problems. So I'm writing the guide that I wish I had when getting started.
I'll publish it on github and post here as more progress is made. I've created three parts. Part One is a non dev focused which explains how each of the moving parts work, what is Ai and what isn't. Part two gets technical and explains the tech stack. Part three is a bonus section that looks at how Ai assistants like google and Siri work and can be improved.
The working title is Automated Agents.
Edit: adding a link for those who want to follow along.
I wanted to record more of my life, but I found that the act of noting it down was too cumbersome. Looking to minimize friction, I created a Telegram bot that saves all messages you send it into a Google Spreadsheet.
Hashtags can be used to split the text into sheets and columns, if so desired. Besides jotting down quick thoughts, this is very handy for short-form journaling such as tracking expenses, workouts, mood, period, weight, diet, etc., with the added bonus of easy charting and summarization from within the spreadsheet. It also supports pictures and other attachments that are uploaded automatically to Google Drive and linked into the spreadsheet.
I'm experimenting to figure out the simplest setup possible that allows me to expose an internal web service (eg localhost:3000) to the public Internet (eg app.example.com).
A million ways to do this, of course, but I'm focused on using wireguard so that eg only my wireguard peers can get access to my local service, and for internal traffic (ie vpn).
At the moment I'm settling on having a simple script that I can run on a host alongside wireguard. The script will function like `wg-quick`, parsing a wireguard config file and handilng the routing stuff behind the scenes, and returning a cleaned up config to be passed to `wg`.
Ideally, the wireguard configs could be generated by some other tool or service, like https://wirehub.org, and automatically fetched and applied to the running wg interface.
So, a one liner on a server with a public IP and the services exposed by your wireguard peers can be accessed via a custom domain name while still respecting internal wireguard routing rules (based on AllowedIPs).
If anyone finds this interesting and wants to chat about it, I would love to! My contact info is in my profile.
This is what ngrok does right? I'm really interested in how this works, I've never looked into it before. Presumably it requires some kind of dynamic DNS updates if you don't have a static IP?
ngrok, localtunnel, pyjamas, the list is endless..
Yeah, many ways to do it, one is running your own DNS server and have wireguard connections use that DNS server.
I want to avoid having DNS issues, so I'm thinking more like this:
1. DNS CNAME *.internal.example.com 123.23.45.67
2. On 123.23.45.67, run wg.
3. On 123.23.45.67 , run nginx. nginx must be in the same network to do geo blocking on Wireguard peer addressess.
4. Create one nginx server config per service to map domain names to upstream servers. Use variables for upstream servers, allow nginx to start even if upstreams unreachable. Add internal locations for custom errors (Forbidden, Unreachable, etc).
5. When connecting to 123.23.45.67 via Wireguard (ie, Peer Endpoint = 123.23.45.67), ensure 123.23.45.67 is in the range defined by Peer AllowedIPs.
https://www.notationer.com/ (a tool to switch across (code) notations, like camelCase, PascalCase, snake-case).
Frontend is mostly in React/Next.js (Some older projects have frontend still in the old ASP.NET MVC and Ajax :-)). Where the backend is needed, it is written in .Net/C# and running on Azure.
My truck. It has sentimental value as my first car. I just gone done with the water pump and I just solved my steering "pop" issue. Next up is a new steering wheel as the old one is gross and sticky. While I'm at it I'm going to install a new turn signal/wiper control lever and fix the turn signal auto-off spring.
I just replaced the rear brake pads and rotors on my truck. PITA, but it's fun sometimes to get outside, have tools in your hands, and work on something tangible.
I have another project lined up for my truck that I'm going to get into soon. I'm going to add a secondary high-current power distribution bus with a couple of bus-bar distribution blocks at convenient points, to make it easier to add accessories. I've already replaced one of the battery terminal screws with an aftermarket one that adds an extended stud and a nut, where you can attach a cable with a ring terminal. And I've got heavy duty copper lugs, some #2 welding cable, a hydraulic crimping tool, and the distribution blocks waiting on a shelf.
Once that's in place, It'll be very convenient for adding ham radio gear, an air compressor, a power inverter and some other things I've been chewing on.
Cool! What's yours? I just found a 1984 Chevy Scottsdale for my kid. It was my first vehicle and now he'll have one too.
It's a sneaky way for me to limit his time tearing around the countryside with friends, because all his money will go towards parts and (if I get my way and put in a much larger engine) gas.
Scams usually start with messages. Predominantly SMS/email sent with links.
ML models exist to determine this, but we are prototyping using Gemma2 2b to utilise its natural language understanding to see if a better firewall could be built, one which users can talk to.
I'm working on Relay [0] -- a collaboration plugin for Obsidian using yjs to provide live cursors and folder sync.
I'm currently thinking about how to robustly Integrate different edit sources like iCloud, obsidian sync, git, and yjs updates. I think it could be cool to create a crdt persistence format that can live alongside markdown files (like note.md + note.md.crdt) to support edit history tracking from multiple users and their devices.
I've been taking a (mental) break from programming and doing a lot more climbing, getting back into my body. After twenty years of boulder/sport/trad/big wall I've drifted into the dark art of rope solo. It's an wild way to climb that really appeals to my engineering brain - lots of deep knowledge about gear and the mountain intertwined with various logistical puzzles. Very rewarding when it works. The latest mission has been optimising my system a lot to start tackling bigger stuff.
Thats great to hear. I'm also working on my body after 2 spinal surgeries. Doing pilates each day, and additionally weight training and cycling on each alternating day.
For a skinny guy who never did any resistance training, this transformed my body to a lean, muscular type in what it seems to me like overnight. I cannot emphasize enough how good that was (and is) for the body. The body at first doesn't like it, but giving it a bit of consistency it learns and yearns for it.
The back pains I had are gone and even if they come back here and there they are much less intense and more manageable. Doing some impromptu exercises for couple of minutes fixes that too.
One of the biggest wtfs I had was when I decided to drill some holes in the wall because I was unable to for months after last surgery. I've grabbed my battery powered drill from the shed and was open mouthed how suddenly light that tool was. I literally checked if I've put the battery on. You don't notice the strength gain that much during normal work in the house, but do something which you know was heavy before and be amazed.
I definitely credit good people on this site for sharing their experiences. At first I thought that only weirdos and meatheads do resistance training and always had a mental image of steroid laden walking muscles. Then I've read that article where was the graph of expected life expectancy for a person taking on training in their forties vs not training and it really did concern me. As a guy in early forties I didn't want to spend the rest of my life with very poor quality of life/mobility. I didn't want to be glued to my computer all day long as it would virtually be the death of me.
I wanted to write a blog post about this journey (which is still short, but memory is fresh;) maybe to help geeks like us to see that being in front of the computer all day long is terrible. It works when you're young, but it absolutely doesn't as you get older. And everyone will get older whatever the younger people think. If I knew stuff I know now about fitness 20 years ago, I'm sure I would be 5 times more healthier, and I wouldn't need any spinal surgeries. But nobody told me anything and I thought "doing stuff" on the computer all day long is fine. "I'll be mobile and skinny even when much older, what else would I need?". Oh boy how wrong was I.
I am currently working on a custom markup language called atex. It's syntax reminds of latex syntax, but with @ instead \ as a special character (very similar to the Lout language, if anyone remembers it). Also, the atex language hasn't any predefined commands. Instead, all commands are defined via schema specified in a separate YAML file. Schema defines commands that can be used and means of "rendering" those commands to different targets (HTML, Tex, Typst, whatever...)
Just today, I finished first working version of the new compiler (https://github.com/ubavic/mint). It is written in Go, and there are lot of things on the TODO list, but it works :)
This is actually the second compiler for the atex. The first one was written in Haskell and compiled fixed document schema. I used it for writing a book on Haskell (https://github.com/ubavic/programming-in-haskell).
I'm currently working on an Art Project stemming from the mixture of cellular automata and neural networks in Rust. The YouTuber Emergent garden has already implemented this idea but I wanted to go much deeper. With different coloring algorithms, neuron types, time travel values, arbitrarily connected networks, crazy activitation functions etc. I just uploaded a very short showcase of it here:
https://defguard.net/ - open-source SSO service built with rust on top of wireguard & OIDC. It's been a lot of work but we're slowly gaining traction and nearing the 1.0 release.
Some of the features:
* OpenID Connect based Identity Provider
* OpenLDAP synchronization - currently supporting users and groups synchronization
* MFA with TOTP, email, WebAuthn / FIDO2, crypto wallets
* wireguard client GUI integrated with OIDC, supporting multiple locations (https://defguard.net/client)
* secure enrollment & onboarding
* yubikey provisioning
Currently we're working on external OIDC providers integration.
65 Words is an anonymous challenge to write 65+ words daily in the language you're learning.
The real focus isn't just on improving your writing but on having the opportunity to think about how you'd express something without the pressure of the moment.
There's no hidden science behind choosing 65 words. I found it’s achievable even on a very busy day.
The goal is to help learners practice daily and build confidence through consistent effort.
I've been working on an electronic music controller designed around seven-limit just intonation that I'm calling the Mosaichord. It's a 4-octave keyboard using a scale with 28 notes per octave. The keys are pressure-sensitive.
Recently I got a DIY strobe tuner I've been working on up and running, which pairs nicely with the Mosaichord.
The Mosaichord works pretty well these days, and I've assembled a bunch of them. Just got a new batch of PCBs yesterday to make more. There's a basic website up, but I haven't yet set up a proper web store.
I have been making YouTube videos about a project I built a couple of years ago. It took a while to find the time and the right approach to documenting it, but I'm finally happy with the results.
I also recently left the Big Tech world after 11 years at Google, so I'm trying to figure out what comes next. (I don't think I can make professional YouTuber pay the bills). If it's not inappropriate to mention here, my resume is at https://hire.masto.me/
Maybe, but I doubt I could make them at a price that would earn me any profit.
I'm hoping at some point someone will comment or message me to say they've built one. I just added the parts list and CAD models to https://github.com/masto/LED-Marquee/tree/main/hardware, and in the next video I'll explain how to build the software.
While owning a printer is still a curse (1), and software continues a cloudward spiral (2) I’ve made solid progress on a local-first indie barefoot app for designing and printing labels. I just shipped a new feature where you can use modern JS to build logic for each label, modifying variables, hiding and showing objects, formatting dates, etc. More at https://label.live/guides/add-logic-with-script-variables
I left my home country for Portugal when I was 18, and was refurbishing iPods for money while living in a hostel. In my free time, I developed a Bluetooth kit, a crude first version which required hours of hand-soldering soldering, CNC milling, etc. to manufacture. While selling these I used the money to prototype and manufacture the second revision.
That's the version you can purchase today, which can now be assembled in about five minutes. This has allowed me to scale the business, and now we're a few employees big, shipping worldwide. A few grey hairs have appeared at 22 but finally I can relax a little.
In my novel (see profile), there are stories within the story as well as simultaneous actions. I wanted to typeset simultaneous events set within a sub-story. Using Markdown, a natural way to do this would be:
> ::: simul
> Simultaneous section 1.
> :::
>
> ::: simul
> Simultaneous section 2.
> :::
Still working on https://exist.io, a platform for quantifying and tracking data about your life. We've been around for about a decade now and seen many competitors come and go — this niche is definitely not at the peak of the hype cycle any more. On the other hand, Apple Watches have normalised tracking body stats like activity (and sleep to a degree), and yet we think there's still so much more that can be done with that data. We're still (always) working on improving our insights and correlations to find better ways to use the data to optimise your health, mood, productivity, and so on. But not with "AI" :)
I've been working on something to handle the "plumbing for your observability". It's a way to collect, stream, and route observability data that can get it from and into a wide variety of systems.
It's not targeted at the 'cloud native' crowd, but rather more at traditional on-prem infra.
It is all manageable via an API and provides a Kafka API for streaming data (RedPanda under the hood).
I had surgery for a torn meniscus repair. I’m on the couch for a least another month.
I’m using this precious expanded free time to compose music, my primary form of artistic expression.
Software-wise, I’m on a platform team for a large company. I’m making some performance enhancements to our http client, and a plug-n-play library for app devs to easily integrate LLMs in their products.
Piling on the "Do the Physical Therapy" train, I had this done 2 years ago, did the PT work and now back 100%. It's a pain (literally and figuratively) but future you will be super happy current you did the work. Look into chair yoga, it will help you keep all of the other bits stretched out. Good luck!
Although I am currently 16 and the summer vacation is coming to an end, I have made big progress during these few weeks. My mission was to build an app with the help of the Flutter framework, and I have succeeded!
The app's website is vid-note.com, if you would like to check it out. The value of the app is that one no longer has to watch endless videos but can find exactly what one needs about a certain topic within the notes that we offer. What is good to know is that most of the notes are centered around business and self-improvement for the reason that we, the founders, are interested in these topics. We understand that many would advise us to let an LLM write the notes, but for this project, we have decided to sway away from the flawed results that LLMs bring with them.
This is the first venture my business partner and I have decided to pursue, and we look forward to finding out if we must pivot or preserve. For those wondering, the app has been approved for external testing on the App Store, but we have decided not to make the jump yet, simply because we still have to handcraft a few notes ;). We predict that a release on the Play Store will soon follow. Cheers.
I was working with the Facebook API for another project, but I was surprised that a website as popular as Facebook had exclusively old API wrappers for it, that all required too much setup, and were cumbersome to write code with.
I wanted a better approach using the standards I have come to appreciate with modern API wrappers, and so I decided to start work on a more intuitive, faster approach for the Facebook API.
My initial launch goal is to allow for programmatic authentication, posting, and commenting, all with one setup command through a CLI.
I am currently working on developing a procedural, interpreted language that is amenable to genetic (evolutionary) programming.
I have decided to shelve my work on architectures that are biologically inspired for now. I was getting reasonable results with spiking neural networks and evolutionary training, but there are so many hyper parameters to think about and how they behave over time is really hard to predict. I was also struggling deeply with how to manage topological concerns like network growth over time.
With interpreted evolutionary programs, the memory access patterns are so much more ideal with the program counter stepping through (mostly) contiguous bytes vs totally insane recurrent spiking neural access patterns. You get so many more generations & candidates evaluated per unit time that it can make previously apparent "dead ends" viable, simply because you don't need to have extreme patience to find out anymore. I am discovering that iteration speed is the most important thing in this arena. The faster you find out how bad a certain parameter adjustment is, the sooner you can get to the good ones.
I am also working on an unrelated contract to integrate some back office banking systems. Not much worth discussing there.
I agree visualizing neural networks is challenging. The flexibility and consequences of network connections is also confusing. If you decide to go down that route again, the best work I've seen by far is the recent one by Stephen Wolfram: https://writings.stephenwolfram.com/2024/08/whats-really-goi...
I'm working on a workout API and native apps (iOS/Android) for Strength Level [1] as another interface for my.strengthlevel.com [2]. Strength Level helps you understand your relative strength - how strong you are for your bodyweight.
Using PHP Slim Framework, MySQL, and vanilla Swift/Java. Prioritizing reliability and efficient sync between local storage and server. Currently finalizing API endpoints and feature matching the web app. Next up is adding a subscription model.
It's challenging as:
- Users want to use the app offline which means we need to sync
- We have to match our existing features on [2] in the first version and evolve the API and database to support reliable sync
- Users want to also track their cardio too e.g. Running Level [3] and Rowing Level [4] but that will have to come out in a future version
almost 1000 comments suggests that this is very popular. doing this monthly would really help, although people may want to repost, so maybe some rule is needed for that. probably marking a post as "REPOST" or "NEW" would help regular readers find new projects.
Adding "REPOST" or "NEW" does nothing to ameliorate the problem of the same comment occurring over and over in search results (particularly hn.algolia.com/?sort=byDate searches).
On the other hand, if each repost consists of just a link to the original post, or a link plus a few words, that would prevent the unpleasantness in search results.
fair point, but i think linking to the original post would make it more difficult to scan though the page, and people would simply not read those posts. might as well ban reposts altogether. so i think if reposts make sense then a tag is better than a link. if links are used they would at least need a short description, so for those that wrote a more detailed post at first, a link with a short desctiption would make sense, but that would benefit from a tag as well.
other tags such as "UPDATE" or "MAJOR UPDATE" could also be considered.
maybe something like this:
Please tag your posts as NEW, REPOST or UPDATE. For reposts and updates try not to repost the whole description but keep it short and link to the previous post instead
We’re building an off the shelf software and ECU that allows you to convert anything to EV (classics and low volume OEMs) with enough tweakable knobs to suit every application.
There are some players in this space already, but we have repurposed some OEM grade hardware that we usually reserve for the big players and so we can offer differentiating features such as DC fast charging, bidirectional/V2X, and things like this to the retrofit market.
My ultimate goal is to bring brain signals to the browser and develop neuro-apps, all in public. I share updates weekly on LinkedIn at https://www.linkedin.com/company/mind-tether
Hi there! We published our book(https://softwareengineeringhandbook.com/) in May 2024, but getting visibility has been a challenge. We've posted on HN before and are currently using Amazon ads, but the results have been underwhelming so far. Any advice or strategies for improving our reach would be greatly appreciated!
I think you need to work on the pitch a bit and probably tone it down. Multiplying the promises under "Skills and Insights You'll Acquire" with "Who This Book Is For" just seem wildly unrealistic for one book and an oversell.
I get the vibe of a "become a 10x developer from scratch and land a rock star job in a month" bootcamp. 2/3 of your audience are likely to have seen enough of this already to be turned off. Meanwhile, perhaps some more junior will be intimidated and say "not yet" by the supposed advanced level (the material is also targeting people 10-20 years ahead in their careers after all...)
If you write another book in the future, I'd suggest splitting it into a couple more focused ones, rather than trying to cover all these dimensions for all seniority levels at the same time.
We are working on building a funding subscription for impact innovation called Marabou (https://marabou.pro/). Our plan is to allow people to subscribe starting from $50 per month and receive equity in innovative startups that benefit either human health (biotech, femtech, healthtech, longevity, drug repurposing) or environment(climate tech, adaptive tech, renewable energy, mobility, agriculture).
Our first campaign to fund Ethan Perlstein's research to find cures for rare diseases officially starts on Sep 9, but we already opened a pre-campaign. This one is a one-time commitment and it's heavily leaning on the charity side.
I have been reading the Reinforcement Learning textbook for the past three years religiously- didn’t get the concepts clear until this week, made it past Chapter 3 finally. The excellent YT lectures by David Silver and Emma Brunskill have been helpful but it took me a long time to ‘get it’.
Next, I will try to build something using RL next but try not to use the Gym/Farama stuff to force myself to learn this from scratch.
I'm working on https://reciperium.com which is a website to host recipes, with the twist that uses a specialised language for recipes, and you can fork recipes from other users and adapt them to your liking.
Not sure if I caught the train but still. My name is Max, a software developer based in Central Europe, and I'm working on my course "Building Command-Line Interface App from Scratch in Go" (this is a working title).
In this course I want to teach how to write CLI application from scratch in Go without any external dependencies (exception: driver for DB).
This is not another one course about how to write CLI app using Cobra or any other command-line builder library. Instead, in my course you're gonna write your own Cobra-like command-line builder library (package) to build command-line interface application from scratch.
Thus, you will know how to build an API for building CLI like Cobra and you can use it later on for your future projects. I will also talk about Command-Line Interface Guidelines showing how to make powerful, useful and easy-to-use command-line application which you can use every day in your workflow.
The course is still work in progress and I don't even have a web page for it. But I do have my newsletter where I share best resources about Go. By subscribing to this newsletter you will be notified once course is ready to launch. Also, everyone who's subscribed to my newsletter and will be in first batch to buy a course gets a 50% discount.
So, if you are interested in this kind of stuff join the waitlist for course in my newsletter on https://kovalevsky.io
P.S. I also have daily base newsletter about Go - Daily Golang. If you subscribe to this newsletter you will also be notified once course is ready to launch you and will get 50% discount to buy the course - https://kovalevsky.io/daily-golang.
I am building my very first computer game, Tungsten Moon, a VR+desktop spaceflight simulator with realistic physics and engineering, inspired by Orbiter, Eagle Lander 3D, and a little bit of Subnautica: https://tungstenmoon.com/
I'm working on a command line billing/invoicing system. I couldn't deal with the ones I've tried (too heavy on UI, too difficult to automate stuff with).
So far I just use it for my company billing, but it's quite delightful for me: My process is to save the services we provided in CSV files (exported from time tracking in org-mode, but you could use any tool that supports exporting), then I run this over it and it creates PDF invoices, and stores all the data (which I query occasionally, e.g. for quarterly VAT payments).
Technologically it's weird, the invoice templates are written in LaTEX, and the code is in Common Lisp.
I don't think about turning it into a business, but I think I'll open source it once it's a bit less messy. For now, I just focus on implementing the features I need (about 1-2 per quarter now).
I kinda doubt it, but if anyone is interested in command line billing tools, I'd be excited to talk about it. Contact info is in my profile.
This is super interesting to me as I have been focused on picking up Emacs and Common Lisp this year for my own freelance and startup workloads. Would you ping me or reply here if you do end up open sourcing this?
I just started working on an engine for turn-based table-top games like Dominion, 7 Wonders, Catan, etc.
During university, I spent some time working on an AI agent to play Dominion, but a very large part of the work was building a way to simulate the game.
The goals are:
- Develop an engine that's efficient enough to use in simulations (for training AI agents or analyzing the game).
- Still emit events that can be used to visualize the current state of the game when real people are playing.
- Create primitives that are easy to distribute across a network for remote players/agents.
I'm a tech exec who misses their physics roots. In my spare time I'm interviewing "materials manufacturers" to come up with a viable startup idea. I'm hoping to build a company that leverages simulation and ML to accelerate material discovery.
Keeping key bits of the idea to myself. If this admittedly vague idea excites you let's find some time to talk.
Senior SWE with Comp Sci + Chemical Engineering background, considering a masters in materials science. If you're ever looking for someone who can bridge those spaces, would love to chat. Email etc in bio.
I'm working on a device that records surgery and uses AI models to analyze the procedure for things like surgical technique, efficiency, instrument usage, blood loss, etc. We're starting with using it for medical education because that doesn't need FDA approval, but eventually it'll be deployed in operating rooms. The goal is to enable every surgeon to perform like the greats (most aren't), improve patient outcomes and ultimately save lives. We just got into Techstars. Aside from that, dad life in suburbia
I am working on a SaaS for real estate agents. Customers get their own tenanted database and web front-end with some fancy front-end tools like geospatial searching to keep customers attracted to said customer's portfolio. I have a paying customer who's been using a 1.0 version for more than a decade now. I don't know whether I got lucky or there's a legitimate market for v2.0 out there. I am building it with boring tech as it's a boring product. I guess I should get back to building it.
I'm developing Weavus (https://weavus.app) as a way to take my events away from facebook. It's a social app (mobile only for now) where events are the main element and the home screen shows your upcoming events. The philosophy is no feed, no ads, no addiction, just your real social life.
The idea is also to merge in a single app many needs of events, like group chats, photo albums, expenses splitting, and I'm planning to add more. I realized groups have similar needs to events, so I added them as well (think small private groups, not massive communities).
All of this is stil very early in development and it definitely needs some more work before a Show HN, but we've found it very useful for weekends and holidays already.
Consider it an open beta, so feel free to give it a try and give me your honest feedback :)
Not in the short term, I'm working alone on this and don't know much about decentralized protocols so I can't work on it yet. Also, Weavus is not a social media: there are no posts, just events, groups and their content. So not sure it's as relevant.
Trying to use Literate Programming to create an OpenSCAD library for a Python-enabled version of OpenSCAD which is able to create DXF and G-code files for creating CNC projects:
I made this bookmarklet, which will speak highlighted text in the browser regardless of platform. It also makes the initial letters of each word bold. This is the code:
javascript:void function(){ javascript:(function(){ var selection = window.getSelection().toString(); if (!selection) { alert("Please select some text on the page."); return; } var encodedSelection = document.createElement("div"); encodedSelection.textContent = selection; var processedContent = encodedSelection.innerHTML.replace(/\n/g, " <br></br> "); var words = processedContent.split(" "); var formattedText = ""; var speechContent = ""; for (var i = 0; i < words.length; i++) { var word = words[i]; var chunkSize = Math.floor(word.length / 3) + 1; var boldPart = "<span style='font-weight:bolder'>" + word.substring(0, chunkSize) + "</span>"; var lightPart = "<span style='font-weight:lighter'>" + word.substring(chunkSize, word.length) + "</span>"; var formattedWord = boldPart + lightPart; if (word.endsWith(".")) { formattedWord += "<span style='color:red'> *</span>"; } formattedText += formattedWord + " "; speechContent += word + " "; } var newWindow = window.open("", "_blank"); newWindow.document.write("<html><head><title>Spoken Content</title></head><body><input type='range' min='0.1' max='10' value='1' step='0.1' id='rate-slider'><p id='content' style='background-color:#EDD1B0;font-size:40;line-height:200%25;font-family:Arial'>"%20+%20formattedText%20+%20"</p></body></html>");%20var%20rateSlider%20=%20newWindow.document.getElementById("rate-slider");%20var%20utterance%20=%20new%20SpeechSynthesisUtterance(speechContent);%20rateSlider.addEventListener("input",%20function()%20{%20utterance.rate%20=%20rateSlider.value;%20window.speechSynthesis.cancel();%20window.speechSynthesis.speak(utterance);%20});%20window.speechSynthesis.speak(utterance);%20})();}();
I'm a software engineer with a passion for pencil portraits (https://bpsagar.github.io/sketch). As an artist, I've often thought that a companion sketching app could help beginners and intermediate artists improve their process.
I'm currently working on an app that provides a step-by-step guide to drawing from a reference photo. My goal is to make the sketching process more accessible and enjoyable for everyone.
As someone who has always been absolutely awful at sketching, this sounds pretty interesting. Is there a mailing list or some other type of notification I can sign up to?
Thanks for expressing your interest. I've registered the domain Portrait.Coach, but it's not live yet. If you're interested in staying updated, please share your contact details using the form: https://tally.so/r/npNGy3
I'll email you as soon as the website is launched!
A data-oriented JavaScript engine called Nova (https://github.com/trynova/nova and https://trynova.dev/), basically trying to answer the question of "what if everything in the JS engine heap was stored in vectors in creation order?" This has some interesting properties; it's basically an ECS kind of engine.
This isn't really very interesting or innovative compared to other comments here, but lately I've decided I want to build a home server to do a few tasks: Jellyfin server, photo server (probably using Immich which looks like a nice clone of Google Photos, where I'm almost out of space), automated device backups, etc., so I've been doing a lot of research into this. I've built computers in the past many times, but I got out of it a while ago so a lot of things have changed.
I looked into pre-built systems like Synology and QNAP, but they seem too proprietary and limited, and somewhat expensive for the hardware they have, so I decided to build my own, but trying to decide on all the HW and SW is a real task: btrfs vs ZFS, Intel vs AMD, RAID levels, etc.
I'm hoping to do something similar, but only use HW that I own or can find at yard/estates sales. I bought a 4Tb WD My Cloud this weekend for $5. Found a 2Tb for $1 several years ago and WD replaced it for free.
Open source presenter software. The idea is anything you want to show on screen, all can be controlled through a single page. Useful for any events like concert, conference, camps, etc. You can also use it for digital signage.
Very early stage right now but I hope to release alpha soon. I'm already using it privately right now but there's a lot to do to make it user friendly.
Working on a SaaS that associates a person/company and their tech stack with psychology profiles. The intent is to better connect job seekers with companies.
In short, if you want a person with Language A experience, but Language B is close enough, my software will provide the signal that the candidate is good enough for a closer look.
Example: PHP and Go devs are likely to have similar approaches to how software should be built. Ruby and Go, perhaps not so much.
In tech, I'm interested in native systems programming, at the moment with C++.
Outside of tech, I do quite a bit of "long" distance running, swimming and e-biking.
When out on a run, it leaves lot of time available for listening on podcasts, mainly technical podcasts.
At the same time, when exercising, I want to publish tidbits I've learned, do projects with raspberry pies, arduinos, make statistics over time series (power consumption, electricity price, internet usage++).
When I'm home, I'm thinking I'll do it later :).
What I'm most curious about, is how people, and I, can stay organized. All information feels like a mess :)
I haven't had any checklist for a couple of months, due to holidays and so on. Earlier I've tried various formats; post-it, small notebooks, notes, paper, trello, notes (iPhone) and keeping it in my mind.
Sometimes the tasks are too large, and will easily take days to complete, which puts me off. Sometimes the tasks are too small, and I do them there and then. Sometimes they are right-sized.
When the tasks are written down, they become 'not important' and 'I can do it later', also they are easier to stow away out of sight :).
Not to mention there are also a couple of hundred open tabs of various stuff I'd like to research... but most of them can be probably be closed.
When it's routine, I've got no problems doing stuff like making dinner, vacuuming, exercising, getting groceries and so on.
But when it comes to projects, and publishing, I feel there is no more juice left.
Thanks Rob. It's been a long time since I wrote this much for a 'human' and not for a 'machine'. I'll try to continue reviving a couple of drafts.
I'd like to work on my writing skills. Some mix of nerdiness and personal Journaling. https://dynomight.net/ is a website that I'm quite inspired by, as it's a interesting mix of personal commentary (https://dynomight.net/advice/) and deep data dives into e.g. homelessness and seed oils.
My main barriers are 1) writing discipline and 2) some obsessive need to link/cite every claim I make. This results in me taking months to write an article, be it due to laziness or constant rabbitholes/google scholar searches about an idea.
Any clue on dedicated disciplined writing time or even considering "reducing" the rigor of writing?
I find when I'm swamped it's because I lack decision making skills. It's a soft skill to keep your workload manageable. There's no substitute to constantly making decisions. And I suspect is the key to being a 10x developer (besides being smart)
Over-researching is probably better than under-researching. To reduce time spent researching, try imposing a deadline. The difficulty will be sticking to it, since nobody but yourself will be breathing down your neck.
I've always found adding custom metrics and monitoring to applications to be a big hassle, so I'm experimenting with one that uses the log stream instead of agents/daemons.
I have one that I'm making open source, as a digital montage system.. think of the "memories" feature on Google photos or iCloud.
I basically want that to be open and free. I'll have an app to easily create albums, and have the ability to connect to any storage or montage service you like. You could even make your own app connect to your own storage and montage service, so it can be completely diy. I have a few write-ups but I'm transferring it to notion at the moment so I won't link it just yet.
My other idea is a "bill hamper / consolidation" service, that I'm doing for my sister. She pays me a flat fee each week and I pay all of her bills for her. Gives her peace of mind and allows her to save some money without stressing on paying random bills
Hello!
I'm working on a low-code API builder based on Flyde, the open-source visual programming language I launched a few months ago (https://news.ycombinator.com/item?id=39628285)
Unlike existing low-code API builders, one can easily eject to a ".flyde" source file and run the API wherever they want without being commercially vendor-locked
I'm one of the people working on Prodigy Reloaded (https://prodigyreloaded.com), recreating the Prodigy server to work with the client software in DOS and Mac. The new server software is written in Elixir.
My part has been to generate news headlines and weather maps using current news and weather information. The most interesting part for me has been deciphering the 1980's era graphics format, NAPLPS, which Prodigy uses and making a library to write files in the format. I treat the file format as a data transformation, taking the NAPLPS file generated so far and appending more commands to the end of it. The commands naturally pipe into each other in an idiomatic way in Elixir.
The Prodigy PC client runs in Dosbox, it's how we do most of our testing. If we, or someone, can get dosbox working in a browser then it's possible. But when it's displayed at vintage computer festivals period-appropriate hardware is used.
Google will review the listing and eventually add it to the map, no further action required by you. Once they do, you'll want to get your customers to write reviews.
I build a computer app to make it easier to do projector sewing. This allows you to take a PDF sewing pattern and use a projector (often mounted on a ceiling, shining down on a cutting mat) to see that pattern at life size. You can then easily cut it out and sew it together.
I do think it has uses beyond sewing; enlarging art, for example, or perhaps painting stage backgrounds, or maybe laying out patterns for building something out of wood.
I'm looking for new ideas for my witch.inc comic strips, it's like IT-Crowd, but for witches and wizards. I kinda ran out of them after I quit my tech-support job. Doing database administration now but it's not as rich in corporate slip-ups
Though I am not actually _working_ on anything concrete(facing analysis paralysis. Also, as a coder, I have no idea how to deal with the sales site) this should turn to a monthly post
Also, Should be a messier version of this as a monthly post where people like you and me post "hey I dont have a project in mind per se, but wanna throw some things out there and maybe hit on a collaboration? Or at minimum idea path validation?
I’m building a travisher [1]. Both wooden body and iron.
I found it amazing the importance of the small details in old woodworking tools and how toolmakers solve problems and improve tools. Particularly in very simple tools.
Working on https://www.shareback.com , team of 2. Been working on it for a few years, trying different ideas. We are trying to build something that we'd be happy to use by ourselves so it's been an interesting journey of self-discovery, research and coding. Also my partner in crime is non-technical and sometimes I kind of forget that people might use software in very different way to how, for example, I would do.
Outside of that I am also building a gym up for myself which I could use without glasses (all apps I tried have tiny buttons or controls that I cant see in the gym)
Re. client side prediction and server reconciliation, may I recommend https://gabrielgambetta.com/client-server-game-architecture....? Even if you don't roll your own solution, it might be worth understanding the theory so you can use these libraries correctly.
My interest in sound design and music production took a surprising turn and I now spend my weekends studying alien vocalization. More specifically, figuring out how to synthesize believable alien vocals and make them expressive and customizable enough to be a tool for anyone who needs that sort of thing for their creative work.
I've been learning how to use Blender. For a long time I've thought that working with things in 3D could be made simpler by having an app where your phone is a sort of 3D cursor, and e.g. for modeling you could use your phone as a sculpting knife by moving it around in the air.
So I spent some time trying to make an app that allows you to do that, but can't shake the thought that such an app would work better as a Blender addon/plugin rather a standalone app. And I also am trying to figure out how people work with 3D currently, to see if such a tool would even be an improvement over existing tools.
Cool! Didn't know about the pen, though I've seen clips of similar tech being used.
I've got a long ways to go with this app though so it's still just a pet project. The position tracking is very suboptimal right now, so I need to switch to a better library. And then I need to start adding actual actions because right now the app is just a glorified object viewer.
I'm insanely flattered that you think it's interesting though. Very motivating to me.
I love this space of hybrid computing. If you want to see some inspiration, look at
- bill buxton at Microsoft research
- Disney research has a lot of really cool stuff too
- anything siggraph
Please keep me up to date one way or another. If you have linkedin, add me, a blog or newsletter? I'll sign up. Discord? Let me know. Just let me know where you're active.
That was cool! Yeah this is exactly the kind of thing I'm working on too. I don't have it that you can draw like he does, but it does let you view an object in a similar way & it can stream data to a server connected over wifi so that you can see mirrored behavior on a desktop app. But it's missing so much still, you can't interact with anything yet.
I believe typescript / js is missing standard library and find it rewarding to work during weekends on those lower level libraries/modules with no/low dependency fanout, publishing them here [0] when they're ready to see the light.
They're driven by what problems I see at work and while playing with other weekend projects so coverage may be asymmetric (some parts are well covered, others that should be are absent <<ie. documentation>>).
I find definition of success as just "working on it" very pleasant.
Evy is an app for collectors, both professional and casual, to help them keep track of their items and share it with the world. Currently there are either overly complex inventory management systems, which are overkill for casual collectors, or generic social platforms, which are "fine" for everything, but great for nothing. My goal with Evy is to fill this gap by offering an easy way for any collector to manage their items.
In short, it is an asset management system tailored specifically for collectors.
Currently working on game development in Godot. We're making Dice'n Goblins, a dungeon-crawler RPG inspired by classics like Etrian Odissey and Wizardry VII, with a cartoon aesthetic similar to Paper Mario. The twist is that you have to collect and use dice to beat the monsters that crawl inside the dungeon.
I am working on an open book Library and a "new" D-Licence.
It is an experiment for the exploration of the free aspect of blockchain storage, where the book library is now permanently hosted on the blockchain, and it's content voted upon by its members.
The licence is based on creative commons, and enforces that the data is decentralized and doesn't need any account, or wallet, to use.
There is also the creation of a new label, the D-Safe label, for a safe experience across generations.
A fun side project - which I have been working on for several years now - rewriting it fully already 10 times - and been restructuring my mental health around it.
I'm a Master's student in Artificial Intelligence at UofT. Right now, I'm adding the "ML (Machine Learning) Streaming" feature to PyMilo. PyMilo is an open-source Python package that provides a transparent, safe, and end-to-end way for users to export pre-trained machine-learning models.
After version 0.9 release, PyMilo became feature-completed with full support of scikit-learn models, now it's time to move on to PyTorch and then Tensorflow. But we decided to add the "ML Streaming" feature before getting into PyTorch, in order to provide an easy way to smoothly stream your ML model. By using the "ML Streaming" feature you can easily deploy your model into the remote server, connect to it from the client side, and choose the working mode, either delegation or local mode, through delegation mode your requests will be relayed to the remote server and you can easily work with your remote model from any devices without any further dependencies, and finally, you can download model for local use.
We will release the 1.0 (tenth) version of PyMilo around Sep 16th, this release will be the first release to have the "ML Streaming" feature with support of REST API, and we will next add other protocols such as Websocket.
Derived a SOTA method that finds the proven optimal solution to the multiperiod blending problem.
This is a problem that every refiner solvea to select raw msterials. They traditionally solve with local optimization methods (LPs) that rely on initial guesses and return suboptimal solutions.
For the same refinery we can find raw material acquisition solutions with 2-5% higher profits.
We made a cloud platform for it and hoping to change the landscape.
I’m building a door. I find that mortises (my frame is mortise and tenon, the tenons were easy on a table saw) are much harder to create if you don’t have a proper plunge router. I’m using some chisels and it works fine but I have eight mortises to make and I’ve finished 1.5 in about the same number of hours. I don’t want to buy a plunge router, because I’m unemployed at the moment, but it would finish the job in minutes. I guess I’ll go get to chiseling! Next door I make will use half laps, so I can use the table saw for both ends of the joinery.
So back in 2021 I built a loyalty platform for a chain of gas stations in the Caribbean. Thought it would be a quick deliverable, then hands off.
It's turning into a full time side gig that's paid off in handful of batches (~$25K and running)
I'm trying to become a loyalty vendor API.
Would really like to work on my own interconnected systems of blogging tools and social media. Kinda a blend between having your own site and myspace? Not really targeting a market or anything like that, just think it would be cool to do for personal use and sharing with family and friends
I went to school with the son of the owner. For about 3 years off and on he'd been asking me about building apps, where he could find good devs, what price to pay.
I'd sent him some info, but never got too involved. Then the pandemic hit, and I found myself with a bit more time to kill, so I listened more to what they needed.
They basically wanted to get a loyalty program going at their gas stations and partner gas stations, but wanted custom rules which their current provider couldn't do with their off the shelf solution.
They also wanted to use one of those old school CC machines that were everywhere before the likes of square showed up. But getting your hands on those units is expensieve.
So I started planning it out. They needed to store customer data, and needed a point of sale device to swipe customer loyalty cards on, which needed to read in customer info, do a deduction, or a store credit top up and print a receipt.
I found a android terminal machine off alibaba, opened up android studio and started hacking away. I hadn't done more than a basic android hello world at this point.
For the backend I used django to get started quickly, and they were very happy to use the django built in admin panel to do stuff as needed, and view transactions.
After I got the devices talking to django, I had to integrate a loyalty vendor they already sourced, and just went along with it, but that was a mistake. The vendor points API doesn't really add much value on what django can already do for me, and my new goal is to become that vendor since I think I can do their product better.
But basically I iterated on it for about a year, launched a little over a year ago, still working out some edge cases and kinks, but they do roughly 25K transactions a month on devices spread out at over 20 stores.
They want to bring it to new islands, so I'm trying to remove some of my duct tape fixes with more stable fixes so that can be smooth sailing come next year.
EDIT: Might get back to writing and update with a much longer post. Draft:
Wow, thanks. This is fascinating and it's also a useful outline or storyline for how these sorts of B2B projects come about. Particularly the part about how their current provider couldn't or just didn't want to provide or develop new services or perhaps just didn't see the customer base developing. Driving a wedge between big enterprise solutions and "move fast and break things" guerilla solutions.
[EDIT: Just wanted to add that I'll definitely be circling back to read the rest as you finish it and the lessons learned.]
I've been working on an exciting project called DataSignal, where I'm developing a sophisticated Named Entity Recognition (NER) and enrichment system. This system is designed to identify and extract key entities from blocks of text, such as names, locations, or organizations, and then provide rich contextual information about them. The goal is to transform raw text into meaningful data that users can easily understand and leverage in various applications.
The real magic of DataSignal lies in how it delivers this contextual information. It can be seamlessly integrated into your workflow through a browser plugin, which instantly highlights and explains entities as you browse the web. For developers, DataSignal can be embedded directly into a webpage via JavaScript, enhancing the content on the page with dynamic insights. Additionally, it offers a REST API, allowing for flexible and customizable integration into other software systems.
Whether you're a developer looking to enhance your application with smart text analysis or someone who wants to streamline research and data gathering, DataSignal is poised to bring a new level of insight and efficiency to text analysis. This project aims to make complex data more accessible and actionable, transforming how we interact with and interpret information.
Bungee is our audio stretching library. It can change music tempo and pitch effortlessly in real time on all common devices and browsers.
Bungee is unique in its quality, performance and controllability. Every media player, DAW, video editor should use something like Bungee for smooth scrubbing and audio slomo.
I know my project is unworthy of upvotes from HN, but I’ve been working (hard) on a simple real estate site https://davegooden.com. No big deal. No real tech. Just lots of work and lots of potential upside.
I'm trying to make it easier to run clubs, associations & organizations with a platform called embolt.app[1].
We're offering online memberships, event management, and a member database packed with features. Membership management is a crowded space, but it's also a low-tech space with lots of sleeping giants not willing to iterate on their product.
It's been a really fun project so far and even more rewarding to see clubs using embolt for their daily operations.
Looks very cool from what I can see. Any plans to have an open / public API? For many associations past a certain size, an integration is going to become an inevitability (syncing member lists between systems is the most common, but also events, i.e. letting people register for an event on your site and then syncing to Zoom for automatic registration and email handling).
Working on Pulselyre, a touch-focused Windows app for producing electronic music live. It doesn't output audio on its own, but it lets you configure various virtual "instruments" on screen that can send MIDI note and control messages to other MIDI devices or VSTs configured to receive MIDI messages. You can record notes and events for each instrument and then loop them over a configured number of beats. Also has some other features to make creating music easier, like saving/loading note sequences, an arpeggiator, receiving input from external MIDI controllers/keyboards, and some other stuff. I've been meaning to record a demo video, but I'm not actually very good at playing or making music myself, so I haven't come up with anything presentable yet. I'm also not really married to the name, but it works for now...
It's built using C# and WPF, and a related project I work on is an open source MVVM framework called UpbeatUI for making WPF apps that behave vaguely like mobile apps. It's for apps that have a main bottom layer and modal popups that float above and can be closed by clicking/touching the background. Pulselyre uses UpbeatUI, and I actually originally extracted UpbeatUI from a much older version of Pulselyre.
https://www.justbeepit.com/ started as a web annotation tool that I hacked together with a friend. Now, it’s a full-fledged visual task management platform with over 2,000 users.
In 2022, I wondered what it would be like if my manager could leave comments directly on the live website for all the edits, rather than sending a bunch of screenshots and videos. That’s how I came up with the idea for a "simple" browser extension that lets you leave comments on any site, anywhere on the screen.
There were several challenges along the way:
Attaching Comments to the Correct HTML Element: I initially struggled with ensuring comments were attached to the right element, as relying on X/Y coordinates would not be responsive. Now, I use a combination of element IDs, classes, outer HTML, and attributes. This approach works correctly 95% of the time. Do you know of any other methods to find the correct element in the DOM?
No Third-Party Libraries: Extension development only allows pure HTML, CSS, and JS, with no external scripts from CDNs. Building a text editor from scratch was one of the most challenging parts.
Real-Time Functionality: Keeping the extension's background page active was tough because it deactivates when not in use, making it almost impossible to maintain a socket connection. I wrote code to wake up the background page and reconnect the socket whenever it goes to sleep.
I stopped all my freelance projects, and now three other people and I work on this tool full-time. We recently became #1 on Product Hunt: JustBeepIt.com
There are still many issues we're working on, such as using it with iFrames or inside scrollable objects, but we're tackling these challenges one by one.
Have been working on a protocol library [1] meant to be a substitute for "JSON-RPC over HTTP/WebSocket". Part of the goal is to more natively support byte-oriented data structures and reduce overheads. It supports a bidirectional HTTP2-style "channel" concept, but without making any assumptions about how they're meant to be used. It also can be configured to require authentication of the client and/or server, but punts secrecy to a lower-level transport protocol. It uses Rust macros to generate traits and wrappers to make the ergonomics fairly easy.
Currently I've built a call/response RPC abstraction with it to use it for another project that I'm working on. Eventually I'd want to add a streaming message system beyond the basic call/response pattern. I also might like to ship some easy setup to bridge between these protocol abstractions and ZMQ, JSON-RPC, be able to behave like inetd, etc.
It's functional enough to use in non-production environments but not completely seriously yet. I'm working out the kinks and improving the DX by using it for that other project. There's still some easy improvements around the byte encoding that I still haven't gotten to yet.
— I’m still curating my newsletter / post series focused on humans and their blogs to help people rediscover that part of the web (https://peopleandblogs.com)
— I’m working on a second newsletter that’s more discovery oriented but still focused on the personal web space
— I’m setting up a private, invite only discourse forum to create a space where people can hang out and connect in a more meaningful way
— I’m working on a new studio site in collaboration with a friend
Open source online backgammon. Backgammon is fairly niche, so there seems to be a lot of opportunity for backgammon software to be developed. I'm taking inspiration from Lichess, having enjoyed playing chess there for years. Check out the server, client and backgammon AI I'm working on at https://bgammon.org
It's fast and easy if you are willing to accept big losses, but slow and hard if you want to save thousands of dollars.
However, it's important not to be too cheap - a friend of mine was stuck with a $150,000 balance on CS.Money because he only wanted to withdraw money in the most optimal way.
I had this realization that, sometimes, people on my team are doing the same tasks in different ways, resulting each of them taking different amount of time. In certain scenarios, there is an optimal way of doing things, and I don't even know about it. Trying to test out if current AI-models are solid enough to do comparisons from a screen recording, or just general suggestions for improving white-collar workflows. I could see an eventual platform that could be built out of it.
It's possible they aren't talking about code at all.
"This team member, and this other team member are both filing a particular kind of report but are doing so in very different ways, is one of them more effecient?"
I like it. Seems like there could be lots of ways to automate the creation of SOP for many businesses. I do wonder how high level of tasks it could pick up on. Though maybe that gets better with a bigger context window.
Hm, if you mean software design patterns, then not exactly. It's probably harder to detect in just coding world, but the easiest one I can think of in our realm "checking current CPU usage of X service". In my current company, there are quite a few ways - logging into AWS -> switching to production environment -> going to ECS -> Looking up the service -> Clicking on metrics, or opening DataDog -> Dashboards -> Service Name.
I understand it's very reductive and simplified, but for every task there are alternative ways to reach the goal, and from my historical experience, everyone does everything differently. Some things can be optimized significantly, if others just knew about the possibilities.
I'm working on Devlog — https://dev.log.xyz — software that I hope will help teams work better by making it more obvious who is doing what. It should help answer questions like "what did Sarah do last week?" or "who worked on frontend performance in the last year?" It's very much in alpha-stage, more a tech demo than a product, and I'm not sure if it will actually succeed. My cofounder just bailed on me but I'm going to keep trying to make it happen. Dogfooding it has made me a better engineer — I write better and clearer PRs — and as a former Senior Engineering Manager, I know it would save me time if I were still managing a larger team. But the setup is onerous (you have to connect your Github so it can analyze your PRs) and the value prop is more vitamin than painkiller unless your team is somewhat disfunctional, and no one likes to admit their team is disfunctional.
Happy to talk more about this with anyone. If you're at all interested in "software to make your team run better" I'd really love to hear what you're having trouble with and how your team runs, maybe I can figure out a way to help you, even if it's not at all related to what I've built already.
Right now it only looks at Github PRs, particularly the descriptions and titles (although I'm also experimenting with the code diff itself), for these reasons:
- Most teams that I'd sell to use Github PRs to write code.
- Looking at PRs, the "common unit of change" of most teams, lets Devlog work for teams regardless of how they end up merging in changes (merge commits vs squash vs rebase... doesn't matter because the PR is what is reviewed and submitted)
- I believe that part of being a great engineer is learning how to describe (a) what your code does, (b) why you're making the change, and the PR title+description is where we as an industry expect you to communicate this information.
- A commit history filled with great PR titles+descriptions is extremely valuable for your team and only becomes more valuable over time, so building tools that analyze this and incentivize you to do a better job of writing good titles+descriptions is good for your whole team.
I'm extremely open to looking at different sources of information (Linear/Jira/Github tickets; the diff itself; ???) if it improves the product. What were you thinking it should look at?
From my perspective as a dev, the question I find myself always asking is "Who's the SME that can I ask about this code that still works here?" So for me the benefit of your project would be using it basically as an intelligent git blame that says who's done what to a particular line/file/module/project, and why. If that SME isn't available anymore, I'd like your tool to help me build context around the issue I'm working on. I'd want to see at least git commit message history and authors included in the available knowledge. If the repo has docs included, even better.
Another reason I would want git history included is because I think it would make your product more valuable quicker. I see your tool as a way to improve the dev culture by promoting the points you mentioned, but it's not clear how long it takes it to start to pay off and be able to answer some of the more historical questions you've suggested. I question, if my team had poor/nonexistent PRs in the past, when would I be able to get any use out of it, and would there always be blindspots around code that didn't have a good PR.
Thank you for taking the time to give me this feedback, really appreciate it. I think Devlog has the information needed to help answer these questions, but right now it doesn't do a particularly good job of making that easy.
> I question, if my team had poor/nonexistent PRs in the past, when would I be able to get any use out of it, and would there always be blindspots around code that didn't have a good PR.
This is a great question and the answer is that with poor/nonexistent PRs and documentation, Devlog cannot help you other than by incentivizing you to do a better job so that it can be useful — so not that helpful. I'd love to find a way to make it better in this kind of "cold start" case, that's part of why I appreciate your feedback about other potential uses. Thank you again.
My pleasure! I've got a few more thoughts that came up.
It would be nice to know more about how I would interface with your product. It's not clear if it's a daily report, if it's a text box where I ask "What did Intern #3 work on last week?", if it's a chronological page, or something else.
Do different people get different answers? Would a CEO get a different answer than an engineer if they asked what the devops team worked on? Basically do you offer different levels of granularity for different user profiles.
Love the project and looking forward to watching it grow!
- Analyzing grocery pricing patterns for 7 Canadian grocers. I took care of the data acquisition. Now it is time to speak with economists / data analysts who can make sense of it. Are you interested? Please get in touch!
- Gathering and archiving all graphical material from the "Printers International Specimen Exchange". The gathering is done, now working on a writeup.
- ... various side quests from the above. Incl. a tutorial on making mobile-friendly imagemaps
- A "vibe shift" has occurred at work. Separately, my own thoughts have been more unkind/uncharitable towards others. Recently I realized that the two are connected. Isn't it funny how the vibe-shift trickled down to affect my own mindset without me even noticing?
- I think that Google is going to fall apart soon. Before they do, they'll cut money-losing boondoggles like, oh, Google Maps, Google Books, Google Groups... Who will step in to save all that data? And what does a post-Google world look like in general, anyways?
- AI is incapable of the kind of impact that is being promised. But we are still going to have a period of ~2 years until corporations realize it (it sounds like AI companies will run out of cash at that point). Until then, we're going to have a lot of layoffs because Mike-the-MBA thinks AI can do people's jobs. What should normal people do? Sit tight and sit out the game for 2 years until things stabilize?
- All of the above is part of a crisis of meaning that people - and esp. tech workers - are having. There's an enormous opportunity for Government jobs to become prestigious again. It used to be a mark of pride to work for the Government. You don't need to burn out and become a woodworker to "do real things": your local city government has plenty of "real things" for you to do in a real community.
Ooo, that is interesting. There was a website that kind of did that, but you could only compare single items at any given time, which kind of defeats the purpose (if it is to save money).
Ideallly I'd like to see something like a personal basket where you can add your recurring stuff (eggs, milk, etc) and compare the subtotal among local grocers. Bonus points if you add in value of store points + credit card bonuses (which might be significant, like 5% for Amex cobalt at a market coded as a grocer VS 1% for non grocer code like Walmart).
I'd like to do something similar to the "personal basket" as a way of comparing "value" at different vendors. Thinking of using $/100g of common sandwich ingredients to demonstrate how much "making the same sandwich" at different stores would cost you. I'm not aiming to use this data for a product/commercial endeavour, though.
I am making the data available for others to use for academic analysis and/or supporting legal action. The large Canadian grocers are known to price-fix and collude, and it would be interesting to see if there are any patterns in their pricing movements. (My contribution is "getting the raw data")
The data was posted publicly online on the merchants' websites.
I started working on my opensource water meter again (https://y-drip.com/). Hardware is basically done so I'm working on the backend website for those that don't want to setup their own server. More technical details here: https://hackaday.io/project/191398-ydrip
Experimenting applying Meta's V-JEPA [0] architecture for representation learning to chess. One of the challenges is that validating if the model is learning useful dynamics of the game, so I'm using it as an excuse to learn some reinforcement learning by using the representations generated by the JEPA model to approximate useful Q-values [1]. This method currently has no search so I'm planning on comparing with this paper [2] which achieves GM level chess without any search. Honestly, Im unsure if the full pipeline is stable enough to even converge, but it's fun experimenting. I'm bad at chess so I really want to make a bot that challenges the best bots on lichess.
Working on a web app to check various metrics on cloud servers (or virtual machines on-prem) and suggest optimizations to reduce cloud waste (and costs). I'm basically automating a good part of the infrastructure maintenance I was doing as a full-stack dev in my previous job.
There are already a lot of services to right-size servers, but I want to focus on small teams that might not have a full-time devops yet and lack the time to check every server manually.
The app will be simple with an educational part, it should be like having access to continuous audits on our infrastructure, just like the CI/CD tools we use on our codebase.
It already helped me at home to reduce the load on my oldest Raspberry Pi by shifting some services to a newer one that was mostly idle and right-sizing some VMs in the Proxmox cluster of the rack in my homelab. Having a green check next to each server name is a good motivation!
Another thing that is scratching my own itch is server discovery (especially in a multi-cloud environment) and integrations because it's just so easy to forget some servers in a cloud when you do a lot of experiments and keep paying for them for months.
The MVP is getting ready to launch, I'm making the landing page right now, and then I hope to find some early adopters to iterate on the app and make it useful to other devs. I'm also thinking about open-sourcing it.
I'm interested in having conversations to discover the pain points of other people in that space, so feel free to reach out to me!
Keurig for microgreens.
A system whereby the user provides power, water, and proprietary seed cartridges to a device to manage scheduling, succession, water, light, and other environmental factors to successfully grow microgreens as easily as possible.
I've found that filing out job applications is rather boring and takes a lot of my time for just a few applications. That's why I'm creating an AI tool to rewrite resumes and automate the jobs application process. I previously created a scraping tool for Glassdoor and LinkedIn, so this a natural next step. List of features:
* Analyzes candidate information
* Examines job descriptions
* Generates unique CVs and cover letters for each job
I'm working on readjusting after getting married. It was a big wedding, and for a whole web of reasons I'd rather not post on a public forum, I was responsible for a lot of the cat-herding leading up to the big day. Best day of my life, and my wife and I are overwhelmed with joy. Yet, now it's been a week and I'm still sitting here thinking "...wait, what the hell was I doing 6 months ago?"
I’m working on a test integration tool. You write your tests declaratively using YAML (ie. setup the environment, databases, etc., load fixtures, make api calls or db calls, and check assertions), so it’s completely language agnostic, which makes it perfect for grey-box/black-box testing. This way of testing also ensures you’re testing functional, business cases, and gives great stability on an exposed interface, because everything runs with real dbs, real queues and so on. From experience, this way of testing hits the perfect balance between reliability, usefulness, maintainability (when done right), and verbosity (at a previous company we replaced the tests on all of our layers with this kind of integration tests and it ended up being the most stable code base and test harness we ever worked on).
I’m currently polishing the logs and reports as much as possible, and then I’ll add support for more tools (Kafka, Rabbit, Nats, etc.). I have tons of features in mind to improve UX, speed, and bring more value to the tests.
Down the line, I want to find a business model for this tool and sell it, but I need to do a lot of thinking on this side since I’ve never done this before.
Yeah I didn’t came up with the idea first, at my work we used an in house solution[1] but it had lots of shortcomings (hard to read output, hard to debug tests, virtually no documentation, and the code was very hard to extend and fix). My plan is to fix all these issues and more :)
Hey guys, at Dispensed are working on Vending as a Service.
For $100 we ship a vending mechanism for you to put on your own custom vending box.
We then provide online marketplace and payment services (Shopify for vending machines) so your customers can use their phones to buy products from your vending machines.
Great alternative for products that don’t justify big $10k vending machines.
I'm not sure I understand what you're actually doing with this. I think that adding photos of the components you're shipping would make the page clearer.
Last week I made an in-browser notes editor using TipTap & InstantDB! (https://owri.netlify.app/) - the idea is, you can keep all the notes locally and publish only the ones you want. Some things I'd like:
- keep track of your writing streak? graphs?
- publish your personal site/blog
- Write one-off HTML websites that're really just docs+links
Feel free to tell me if there's anything you want!!
New idea:
So currently I'm trying to learn leetcodes, but thinking of alternative/structured ways to do it. What if there's an AI-powered book that's basically like your personal Wikipedia into anything?
You start with a topic, it generates you outlines/courses -- and what'd be great is if this was social, like you can see what everyone else is learning about, what nodes they're expanding...
My current project is TrendBowl (https://www.trendbowl.app) which summarises trending entities across multiple sources on the web, including Mastodon and Wikipedia. On a 24 hour basis you can see which terms are relevant in different languages and sites.
It is implemented in Go, Python and JS, with support from Postgres, TimescaleDB and Kubernetes.
Why no Twitter/X? I believe they have an API that you can use, but they've locked down scraping pretty hard. The API should be cheap however, if you're only scraping trending once per day.
Originally I had Twitter support, but they changed the API around and it ended up not working any more, see the project blog for more info: https://blog.trendbowl.app/
The Pro subscription to the API now seems to be $5000 per month!
Currently I'm working on file compression program in Rust. Nothing too fancy, it just use common algorithms (LZ77, LZ78, etc.)
The only difference here is that the program will switch on the fly between different algorithms depending on which one that can compress file smaller.
It can compress 1 GB file (enwik9) down to around 230 MB. Pretty good I guess for something that I worked in my spare time.
I'm not publishing it yet, since I'm still experimenting with it a lot.
Figuring out how to add game boxart and screenshots to my recently-relaunched MIDI and chiptune player: https://pixeltune.org/
It's a fun challenge, because I need to figure out how to reliably determine the matching game for files like /pub/johnny/midi/games/sc2kmenu.mid, and also to show them in a way that fits nicely in the UI.
I'm working on a serialization format, somewhat based on apache arrow but row-based. Includes end to end typescript type safety, and significantly faster than JSON in serialization and deserialization. Seems to do fairly well in the native version I threw together, but more of a javascript/typescript operation for now.
https://github.com/Trint-ai/TrintAI I've been working on integrating different opensource tools and ML models to create TrintAI. TrintAi is a powerful open-source tool for transcribe and understand speech with multiple capabilities.
This is an open-source alternative to paid services like:
AssemblyAI
Deepgram
Gladia
Google Cloud
Microsoft Azure
RevAI
With trintAI you can build your own custom workflows & integration for speech-to-text transcription.
The main features are:
- Speech-to-Text Transcription: Converts audio files into accurate, readable text in real-time.
- Summarization: Provides concise summaries of long audio files or transcripts. This feature extracts the most important information and key points from the text, allowing you to quickly understand the main takeaways from meetings, calls, or any extended audio content.
- Sentiment Analysis: Detects emotions within the transcribed text.
- Language Identification: Detects the language spoken in the audio file and can transcribe in multiple languages.
- Diarization: Identify and distinguish between different speakers within an audio recording.
Give it a try! it is open-source. Your feedback is very appreciated!
I am building a belt tension meter to attach my z-belt idlers to on my voron 2.4.
I'm still contemplating what I want the final form to look like, but currently it's a 5 kg load cell and a floating idler and an rp2040 reading values via hx711.
Navigation using photos.
Did not like existing apps, so made an app that lets you match a photo to an (OpenStreetMap.org) background map, and shows your position on that photo during your trip.
Works great with physical maps, screenshots of maps or downloaded pictures of maps.
NavigateAnyMap.eu (Android only, sorry!)
Free as in beer, will tell you is uses ads, but have not added them yet (and will make a paid, non-ads version first).
An automated dubbed translation service. Supports translation into English, French, Spanish, German, Portuguese, Chinese, and Japanese. The other options on the market often have jarring audio artifacts and glitches, which Audiomatic avoids.
My friends and I are currently working on improving the voice quality and adding new features.
I have been working on creating a hosting project where I can provide access to lots of private frontends (invidious, redlib, etc) and different services (vaultwarden, nextcloud, etc), etc. I have been setting up glusterFS (a distributed nfs) which I can use in Docker swarm to share the volumes for different nodes. I would wanna work on writing a distributed nfs once I am done with this small project
I’m building an integration platform, think Zapier or Workato.
We’ve been hand coding full-service integration workflows for 2+ years on a large B2B agency model, and in parallel have been building a platform using what we’ve learned to build better abstractions.
We’re at an interesting point where almost all new workflows are being built on the platform, and we’re serving over 100 clients.
We’re prepping for public release in the next few months.
Sorry, just saw! I'm working on something very similar in a very niche space, also trying to take lessons learned and productize a more general solution / platform based on past experiences.
It's a steep cliff between one-off integrations and a general platform with workflow/trigger/action type functionality (assuming you do anything like that, since you mentioned Zapier).
In my experience the hardest part of a general solution is all of the one-off custom business logic that each client seems to have, even when integrating the same two systems. There's custom fields, which in one system could be represented as a special object and in another is just another column in a database. There's tons of if-else logic. Etc.
When doing something like field mapping, did you end up creating an internal/intermediary data model to map everything to, or did you end up creating System-A-to-System-B type mappings? How did you end up handling the special cases; is there a sort of scripting system for special and conditional logic or do you have something like a set of predefined transformations / conditionals that can be applied to the data? I'm sure a lot of this could be handled by a workflow/trigger/action system if one were willing and able to go fine-grained enough on the individual steps.
Other than that, what would you say your biggest hurdle has been in getting this thing off the ground and running smoothly without too much babysitting on each individual integration?
Sorry if this is too much or if I'm asking questions about your secret sauce, feel free to ignore, or email me at my username at gmail.
Trying to innovate digital language learning by building actual learning games, not just gamified apps.
Currently experimenting with a game build around the old classroom method of Total Physical Response — you can see a very early prototype here: https://kolja-sam.itch.io/the-tpr-game-gala
I wanted to learn more about htmx so I created a very dumb project https://mettag.ulry.app that fetches meta tags of passed urls and displays them nicely in the UI (also provides APIs because why not?). Still have to finish it, especially the err handling part, but it turns out I quite like htmx!
• AWS Resource Explorer - a lighter-weight version of the AWS console where everything is just a sortable/filterable/searchable table.
• Access Denied Debugger - paste an "AccessDenied" message and get back a stack-trace style UI showing all the resources involved, reason for the error (e.g., which policy is missing a permission), recent changes via CloudTrail, etc.
• AWS Organizations / SCP Viewer - generates a tree-diagram style UI showing all your AWS accounts, which policies apply to them, etc.
Still working on merging these into a cohesive application (mostly just been scratching my own itches so far). I'm trying to consider privacy/security carefully, so everything is client-side, using the AWS JavaScript SDK, and creds/data are only stored locally.
This is awesome, I work at AWS and might use this when the console load times get on my last nerve and I just want to check some IAM policies. The Access Denied debugger sounds like a massive timesaver too.
Oh that’s so nice to hear! It’s quite an early alpha, but I’ve working to expand support for more resource types and details. Feedback is very welcomed.
We are working on https://aide.dev/ an editor forked from VSCode which allows you to utilize LLMs to speed up coding.
Our approach is a bit different since unlike making a better copilot or a chat experience we are building workflows which encourage engineers to work alongside AI and not just offload tasks to AI.
oh I could write a book on this lol, lots of limitations. One of the core reasons being: extensions by design are super limited in the functionality they can provide.
We did have an extension which worked okay while we were still learning the ropes but after 2 months developing it we realized that we didn't have access to many of the APIs which we wanted to tinker with or change the UX dynamically!
This made us look for 2 options:
- either move out of the editor and become a cloud blackbox AI engineer (not a big fan, I don't want to spend my time just reviewing code)
- or own the editor where people code
We chose the editor route and didn't look back after. Over that time we have changed the UX of the editor completely can better play into the APIs which are hidden in VSCode and really build something which I personally could use daily :)
cursor is cool ngl, I do find it missing for the use-cases I would want help. Honestly I don't use LLMs a lot daily (copilot is distracting, chat is ¯\(°_o)/¯ I know the codebase pretty well)
The kind of help I personally want is not something cursor supports .... so the best way to bring our own vision to life is by taking a step towards building and editor and thats what we are doing
I'm scratching an itch. I'm a huge fan of system monitoring and alerting when they go down. It allows the engineers in charge of the production systems to sleep well knowing that things are working.
There's a gap for email monitoring. How many times did you have to awkwardly explain why the notification emails in your app stopped sending a month ago? Or that time when the weekly summary report did no go out on Friday, and on Monday you were chewed out by the boss? Or when Jerry went on vacation and neglected to send the weekly marketing campaign email.With
[I'm building https://wasitsent.com - you get an alert when your emails *stop* sending. It's like an uptime monitor for email sending. If something breaks your email sending process, we'll let you know.
If you would like to see such a service, or have feedback (positive AND negative) about it, please let me know. If you heard of other companies offering something like this, I'd love to know as well.
A long time ago I built a Nagios script for verifying email end to end. It would trigger a test email to be sent from the app to a known mailbox and measure the time to receipt in that mailbox. Simple but extremely effective, and several times it caught a failing in the mail stack somewhere.
It also got used to validate enquires@/hello@/contact forms were delivering, especially to set small business owners minds at rest.
We use a similar email monitoring tool from https://ert.digitconsulting.net/ It integrates with our RMM. So there is definitely a demand for a service like this. But I would expect that paying customers will want it to integrate with a monitoring platform, more than just a message on slack/teams
https://blr.today - Working on an open-source stack to curate events happening in Bangalore. It curates events from multiple sources, cleans them up, then curates them further by tagging them nicely, and making all event data available as calendars you can subscribe to.
My Core-thesis[0] on why I think this is worth building:
1. The vast majority of events in aggregator websites are low-quality, and often filled with spam. The aggregators make money by listing lots of events.
2. Small venues hosting cool events will not always publicize them on the aggregator platforms.
3. The best UX for event discovery is your existing calendar app.
4. Events are highly structured data - but this is often not captured.
I've been wanting to build this for almost 6 years now, finally getting around to it.
We’re working on our contract management system for dev and design agencies. Been in beta for a couple months, working as always on that elusive product-market fit.
Besides that our team is always working on cool stuff for our clients. Lots of interesting work these days “in” AI, plenty of cool stuff going on in health tech.
Got something cool you need devs for? Always happy to talk shop — wyatt at apsis dot io.
So I have started creating a new keyboard layout based on Dvorak; but more a text generation and expansion engine, with snippets or abbreviations in vim.
Initially also on Emacs as a minor mode, the concept is to write in a shorthand of greek letters, mapped like to a Dvorak distribution, that stand there as a code, as variables or latent values, expanded upon a key combination to one of several languages: German, Hebrew or Yiddish. Text can stay code-mixed indefinitely. The greek letters are a preform, condensed by layers of translation and transliteration, a hybrid con, and the keyboard layout itself, dynamic and not static, is dual: for every greek letter in the layout correpsonds a hebrew letter or diphtong or diacritic mark etc. It is a merge and disambiguation of both Dvorak/Programmer Dvorak and Yiddish Klal.
I really don't know if it may seem useful to someone, but I've been itching writing a bit jargon in a system like this for some weeks.
I'm finally frustrated enough by the state of coding that I'm developing my own coding environment and language. The leading question is: What if UI was first-class while programming?
Happy to elaborate if anyone is interested, I will also write about it on my blog (https://watwa.re) at some point
A search of your site doesn't have a match for "Literate Programming" --- I've found it a benison when developing, esp. in that it allows me to review code which already exists, and to create a structural index while writing which allows a quick/easy check if a given module already exists in a context where it should --- managed to make a system for this w/ a bit of help from generous folks on tex.stackexchange:
(probably because of reading Herman Hesse's _The Glass Bead Game_ (originally published as _Magister Ludi_)
I'd give a lot to have a programming system for METAPOST/FONT (or some other graphical language) which both allowed drawing in the view and then editing the textual representation of the code causing the view to update (yes, I should use http://tikzedt.org/ )
Pardon the delayed response, I'm not sure if this will still reach you but I'll type it nonetheless.
I unfortunately am not very fast at reading Tex so I could not quite grasp it. My main touch points with literate programming were probably iPython Notebooks and ObservableHQ. I like interspersing code with (rich) text, and starting with a top-down view of a codebase is imo also the way to go.
I'm less sure about the macro bits, I have yet to see a language where it makes for sane means of abstraction. It certainly solves for a hole in our meta-programming design space, but I don't think we've found the right hammer for that nail yet.
From the ones I know of, Pharo is probably closest to my vision, though it is still very far from it, being brutally OOP (the obtuse kind, for me at least) and it still puts a wall between code and visualization. A wall which I intend to tear down.
Thanks for the Hesse recommendation, first time I'm getting one in my native german on this site.
Tried to rewrite the Javascript 3D library three.js to make it smaller and more specialized for my game. Today I’m ending that and instead starting to write a 3D engine from scratch with the knowledge I learned from this rewrite process. Less object orientation and code organization, in order to get to the core of what happens in the machine (the phone/computer). There is enough complexity in the OS+browser to get through to use the device’s full power, so I at least try to remove all complexity possible from the code. Make it use less memory, download time and CPU time, and also make it go through the Javascript parser and bytecode converter as smoothly as possible. Then minimize the number of draw calls (that’s often what takes most time in the GPU) so I can render more varied graphics at the same speed (if less time is spent on sending data back and forth to the GPU, more time can be spent doing cool stuff in the shader code, I guess?).
I'm working on a little mobile game which surrounds locating and collecting VW beetles 'spots'.
I've been spending a lot of time in México over the last year and have been working on a photography project surrounding the VW Beetles (Vochos as they call them) you'll find all over the country. I'm inspired by the range of character of the Vochos and how they've become ingrained in the country's cultural landscape.
I realized I had a sort of hyper-niche problem where I was locating Vochos I'd like to shoot when I wasn't carrying my camera. I was getting annoyed with using Google Maps to list locations I'd like to return and shoot so I spent a weekend putting the app together and turning it into a game.
After using the app for a few weeks I've realized the spotting or hunting of the cars is the part I enjoy the most about my project and when I have the time I'd like to publicize it so others who find themselves in Vocho dense area can check it out if they'd like!
Sorry, I'm showing my age. When I was a boy it was common to play "slug bug". Sighting a Volkswagen Beetle was an excuse to punch your friend/sibling in the shoulder without retribution.
In the best case, the developers of e-commerce website create unit-, integration-, and functional tests to make sure, they do not break existing functionality.
But a website is not only code. Its also data and configuration.
Since these types of configuration (e.g. prices, legal texts, bonusses) can change quite often and can get quite complex from the business perspective, it does not make sense to have developers create automatically executable tests to make sure, the website is configured as it should be.
And its often not the developers who change the configuration. Maybe its a product manager, a marketeer or the ceo.
These people often do not know if they configured something correctly. In additiom, they often don't get any notification, if an process, that is important for them, breaks due to a misconfiguration or some bug in the code (not every code base has a test coverage of 100%).
So, I am creating a No-Code Black Box E2E Monitoring Tool, that the process owners/configurators can use to regularly check, if everything is working fine.
Mostly just learning Finnish. I tried to get started on the Georgia Tech online master's in CS, but between that, a quite stressful full-time job, and a happy pregnant wife, I just found it was stretching my nerves too thin.
I'll come back to it once I'm farther along in my language learning journey and want to take my foot off the gas pedal a bit.
I need help finding a full time job in Boston so I can move there.. Facing a catch 22 with apartment applications. I have a tooth brush and will scrub makefiles for a living wage, I'm not picky! (resume in profile)
Parallel streaming video ingestion of live events, chunking up the video, sending it to Gemini Pro to get get the context/narrative/transcription/sentiment, and alerting for various things.
I have some side projects that I work on alongside my regular job.
- TheFile.Ninja, which is a file manager with the Everything indexer built-in for extremely fast information retrieval. This allows you to quickly run file queries, and these queries can then be saved or even added as folders in the file system. When you open the folder, it automatically fills with the contents from the query. I have also created a service with LLM/AI that translates plain text into Everything's query language. This enables you to build very complex queries directly from plain text. For example, you can ask if any directories contain a certain file, and if this file is found and contains specific text, it will be displayed. You can learn more about the project at: https://thefile.ninja or watch https://youtu.be/JREufgkf5pk.
This summer, I also built three smaller games (not fully finished but almost):
- ThrustMe!, a space/underwater cave exploration game (https://youtu.be/M0d7CSpEJ1E). It works on Android, Web, Windows, and possibly soon on iOS.
- MergeQuanta, a Tetris-like game where you merge matching blocks (color + shape) to make them disappear, taking surrounding blocks with the same shape along with them. There are also cement blocks and bombs. The game works well with touch and stylus but also on regular computers (https://youtu.be/VXvpzhi8ySE).
I am adding some more details about the projects.. The caves in the ThrustMe levels are fully procedural, every time you start the game you will get a new cave. I'll drop a link to a playable alpha-build (with bugs and it is untweaked so wait for the real release) on the web: https://thefile.ninja/thrust.htm
About QuantaMerge, this game is actually using a Physics engine for the falling blocks, it is not a regular Tetris clone..
Total Commander was the original inspiration for thefile.ninja, but over time the context has moved more and more. The program should be easy to handle if you have used total commander., same keys, etc. However, the program has a much more modern interface and it is above all much faster in searches. It is so fast in searches that you can use searches as your own folders in the program.
I was working on reading the comments on this post, and got tired of scrolling with the mousepad. I had not used any of the LLMs in a few weeks now and wanted to see how they fared for writing a web extension to let me browse hacker news via arrow keys and the "prev" and "next" comment buttons. My goal was to write 0 lines of code myself and rely solely on copy-pasting(successful in my eyes). ChatGPT 4o disappointed me on getting a final product, but Claude Sonnet blew me away enough to push me to pull out my credit card to subscribe to a pro plan.
Building some personal finance tools. I don't like how many personal finance sites seem to be built around hawking credit cards and insurance rather than focused on helping people build a budget and get out of debt.
Aside from bog-standard machine learning for work purposes -- novels! (Because I apparently can't restrict myself to just one.)
You can see some of my world-noodling about aliens that don't need to breathe at: https://rznicolet.com
I've been plotting a series of hard sci-fi novels entirely from the point of view of very not-humanoid aliens. In this case, hard SF = all real physics except for FTL. First manuscript complete. Humans not included in the main series, except as passing footnotes. Let's just say that when I originally started, the pandemic had me feeling a touch misanthropic. We'll see when/how I actually publish these.
I write fantasy stuff, too, though since the blog is relatively new, I haven't started adding that in yet. I'm currently wrapping a sequel to a finished fantasy work (again, publication approach TBD) before switching back to relatively hard sci-fi.
Spec generation from request logs, automatic schema generation and validation, test generation (eventually), totally offline, no accounts or cloud sync necessary!
Been taking longer than I hoped but should be released soon (next week or two)
I'm working on marketing my video course, that meant to help elderly people to use their android smartphones. TBH i struggle a lot, as i'm a software engineer not a markering person:-)
There are many out there, who are intimidated by trivial things, such as sending a text, or even scrolling or switching between apps (what is even an app?).
So this course is ready made just for our parents. Or just for the ones who have no patience or capacity to teach these things step by step to their loved ones.
There's lots of tools in this space, but the key features of this is it lets you easily check the contrast of any color pair (not just against white, so you can check e.g. your text colors will contrast on off-white shades), it's for creating a full palette of colors vs a handful of brand colors (you always end up needing lighter/darker variants for things like borders and backgrounds), and you can alter how the hue/saturation/lightness varies across a whole swatch of colors with a few clicks (being able to visualise these curves also makes picking new colors really easy).
Feel free to reach out if you think this might be useful to you!
This is great! One feature that I use ChatGPT for is “give me the closest foreground color to X that meets AAA accessibility on Y background color”. ChatGPT that calculates it with a python script. Has been super useful. Maybe it could work on your site?
Yeah, that's the kind of use case I want to make easy!
One of the tricky parts when working with ~100 colors is how you can easily tell the tool which pairs should always contrast so it can check them automatically, instead of you having to keep selecting pairs to check.
It's simple right now though to make all colors of the same grade have the same lightness so you get easy to remember contrast guarantees e.g. green-700 will have WCAG AA contrast on grey-100, and the same for any 700 (or above) vs 100 (or below) pairing.
I was thinking of doing something like this. My first idea was the parent manages a sort of simulator (farming, civ, city, idk) and the child is a little chaos agent that wrecks havoc.
But you might get the reverse, parent frustrated kid bored
It is a good idea. I can't have my kid play poe and I am checking out too quickly when I let her play pattern matching games. I personally think this is a niche in Steam that has not been 'occupied' well yet.
I am working on a simple image-hosting service. Imagine Imgur, but you can backlink. Additionally, you can resize, crop, and optimize images on the fly. Here's a sample PNG being served as WEBP[1] reduced from 1.5MB to 75KB (no visible reduction in quality, original image [2])
Not a sexy idea, of course. However, I couldn't find any solution that fit my needs for my SSG blog. Services like Cloudinary, and Imagekit are too developer-focused, and popular image sharing websites like Imgur don't allow backlinking.
Unsurprisingly, I don’t see many hardware projects here.
If anyone’s looking to share their story, I’d recommend IndieHackers, but for hardware-focused startups/projects, check out howtoware.co. (I’m not affiliated with the site’s owner in any way, but I really like the concept and want to help spread the word to my fellow hardwarers).
My hardware project is the 100 year computer. I believe the next phase of computing will be focused on maintainability and longevity. This is different from the development phase of computing that currently surrounds us.
The next wave of interesting hardware development will be focused on form factor. Devices of the future could be any shape or size.
The first version of this device I have completed but not fully documented is an 8 unit mixed use apartment building. I converted each of the 7 classrooms into studio apartment the 8th unit is my development studio.
How can a building be a computer? My perspective holds that a building is a machine that needs a differential engine to optimize the resource usage of the occupants. This perspective with the idea that some circuits should larger not smaller to aid operational understanding.
The building is done enough and I need a break so I think my next form factor will be mobile sensor network and focused on individuals and small groups.
I think developing for a sailboat will be a good complement to the building. The both serve the same purpose but one is extremely resource constrained and in a harsh environment.
Currently identifying all the current sailboat tech for a refit. The devices I’m looking at range from RADAR to sine wave inverters. A very rich tech environment. Looking forward the fall.
Cool, this look great, thanks for sharing. Scanning the front page, I actually think the maker of this [0] is active here on HN and [1] also made it here before.
I'm working on a paid Ruby gem for bespoke Stripe subscription integrations. It cuts down integration time from +6 months to less than a day, and handles many Stripe API pitfalls.
Stripe has a "developer friendly" and "easy to use" reputation. But, since 2019 the EU regulation has stepped in and made things more complex with 2nd factor authentication for payments (3DS/SCA). This made payment integrations way more complex. Integrating Stripe subscriptions now easily takes 6 man-months (I'm being very optimistic here).
Also, there are some basic scenarios that are hard to get right:
- Creating a paid subscription and ensure a customer always has a card on file (this one is almost impossible).
- Upgrading a free ($0) plan to a paid one.
- Upgrading paid plans, eg $10 to $100/month and ensuring immediate payment.
- Guarantee customer Tax location, keep the flow simple.
I made a video analyzing Stripe integration of a successful SaaS company (+$100M valuation). I first paid $30, then upgraded to their highest $2000/month plan - for free!
The gem I'm working on is intended to be used with Ruby on Rails apps. It covers all the above mentioned hard cases, and irons out many more Stripe API kinks. And yes, it handles the basic-but-impossible "ensure customer always has card on file if they start paid subscription".
After buying the gem you can hand-off the task to the junior developer (it's that simple). They follow the integration tutorial: follow the steps for Stripe dashboard config, do the local env config, done.
The product (Ruby gem itself) is ready. I'm now working on the web app, tutorials etc. If anyone wants an early, guided access, please email me - contact is in the profile.
Interesting. I built a recurring billing startup and was always wondering how Stripe dealt with those edge cases...
As a note, I watched the video, the company name does become visible at least twice, once from 6:33 in the header pane and URL and also at the very beginning (Company Insights pane -> below "Powered by AI" the AI repeats the company name)
Interesting idea, getting Stripe to work perfectly in production for a global product is not as easy as it was, it really takes a while to fix all edge cases.
Sometimes it's not about the "edge cases" and handling Stripe quirks. The main design consideration for global Stripe integrations is SCA/3DS or "payment confirmations".
If you implement something, then add SCA/3DS "payment confirmations" later on, I think it's impossible to fix that.
The main pain point I remember from my biggest integration was getting all the webhook events right. There was also a thing about the order in which the events were received sometimes.
I am trying to get better at trading so I thought it would be nice chance to work on my coding skills. I wondered if financial indicators actually makes money for a given period.
Yes, there is a bug somewhere which I could not find so far… it is making me crazy. Yet, I will find it at some point. Other than that, I am still improving the site!
I'm working on a library and accompanying notebooks to process secondary school exam results (UK). My father has been doing this for years for his trust and has a fantastic spreadsheet that spits out loads of data for each school.
I've recently taken a new job working closely with some data scientists so wanted to skill up a little. So far I've got an elixir library which can calculate Attainment 8 and Progress 8. They're formulas provided by the DfE to understand individual student progress, and the schools ability to progress students.
It's just a bit of fun, and is letting me experiment with some of the various elixir data libraries and livebook. At the end of it, his trust is going to plug in some anonymous data and see if they can get more insights than this mega spreadsheet.
It's also nice to be doing a side project which isn't aimed at being a business.
Desperately applying to product design jobs as unemployment stretches into a fifth month.
Keeping myself marginally sane by building a custom browsing experience for my “graphic design reconstructed from dreams” project: https://dreambrief.presteign.com/
These days, I've been working on a small tool to run GUIX as an unprivileged user, so that you can use GUIX profiles as a dev environment builder. I didn’t do any market research, but AFAIK there are similar tools using Nix. Honestly, this is just a toy project for fun. I had GUIX on one hand and Zig on the other, so I figured, why not try something new using both?
Why not Nix? I just don’t like the Nix language. Plus, it seems like they ended up reinventing part of Lisp with their Flakes, which made me even more convinced in using GUIX. The GUIX folks didn't disappoint me, as they utilized the full power of Scheme to design a (internal) DSL that is clean and almost reads like JSON/YAML. I feel quite confident that it’s a better language for building devenvs for most people. The only downside is that it can get really verbose, but, hey, at least you have full control.
One thing is that I was referencing the old popular quote on Lisp: "Whoever does not understand Lisp is doomed to reinvent it," so there’s a little bit of exaggeration there.
Still, it’s true that Nix Flakes is a kind of dynamically-typed environment that lets you compose various constructs in Nix in a uniform way. But, let's be honset, we have had duck typing forever, and Lisp is a dynamic language from 1960s. Reading through forum posts (as I never went deep with Flakes), Nix really needs a fundamentally better language, or I have a strong feeling that the project will just keep reinventing language features one by one.
Hey HN! I'm working on Voop - a b2b phone network, leveraging eSIM & web-based UI we're turning company mobile connectivity into a SaaS. Has been really fun as an engineer getting into the telco stack - networks, switches and traffic routing take me back to university - and commercially it's been fun negotiating with the mobile networks to get wholesale access. Launching later this year in the UK and US next year, can't wait to get it off the ground.
One early achievement was buying voop.com - a very nice man in Arizona had held it since the 90s, and despite being retracted on whois, i worked out how to contact him via his custom ssl certificate on the empty holding site parked at the root domain. After a few emails and follow ups to get a response, I bought it for a not-cheap-but-not-bank-breaking amount.
Currently focusing on building native apps using React Native. Finding limitations with the platform and considering dropping the whole thing and start building with Swift and Kotlin instead.
2) You listen and read and try to understand the meaning with the help of translations and the definitions in context of the words in the sentence. (You can click each word to get a definition in context)
3) Over time you mark words as known
That's basically it. You learn with context using content you are interested in. I'm working on adding a SRS to the app and a feature to improve your pronunciation using the language shadowing technique.
And as mentioned, building native iOS and Android apps.
FHIR is a data standard for sharing healthcare data. It's in use all around the world by healthcare, health insurance and medtech companies as well as researchers working with healthcare data.
Vanya grew out of my own frustrations as a developer working with FHIR. I wanted an easier way to see the data behind the APIs. To say Postman is not ideal for working with this data is a major understatement.
We're a few months away from a commercial release, and have an MVP available for download right now. We're very much building in public and have a strong user base already all around the world.
Just finished a Sunday morning release of the latest version. Looking forward to the next 6 months!
I'm putting the finishing touches on the major engine pieces for my NES game Tactus. (https://zeta0134.itch.io/tactus) The level generation is all working, there's a new raster effect system that can do nifty screen-slides and distortions, the dynamic lighting is more performant, and I'm at the point where it's time to shift gears and start making a bunch of content.
The one zone that's in the game is looking more polished than ever, but it's not going to be much of a game if there are only 4 levels. Next month is all about stage art, enemy design, and items to flesh out the new economy systems. Hopefully it all comes together, we'll see. It's a ton of pixel art, and I'm still learning my way around that craft.
I'm writing a new property-testing library for TypeScript. [1]
I wanted a way to define new Arbitraries that's easier than working with .map(), .chain(), .oneOf(), and other combinators that require you to think in terms of sets. It has those combinators, but you can also write code to randomly generate one value at a time. It uses an approach to shrinking that's inspired by Hypothesis.
Along the way, I ended up adding a Domain subclass, which also does validation like Zod. Not sure where I'm going with that, but it's also useful for generating unique keys.
(The documentation isn't done and doesn't explain what's interesting about it. Caveat: it's Deno-only, and will probably stay that way unless someone wants to help.)
I'm working on planning next year's Square Foot Garden and building some tools to help me figure out staggering and succession planting--trying to be self-sufficient this year. I'm also working on a cookbook that focuses on "composable" meals and minimizing kitchen waste.
I got tired of all the pop-up’s, long stories and accounts needed on cooking sites so I’m building a web app that helps you organize all your favorite online recipes in one place.
It also helps you discover new recipes that other users have saved.
Apart from working on my start up, and open source project https://github.com/laudspeaker/laudspeaker (open source firebase cloud messaging) I've been writing a short sci fi story, sort of like a techno thriller inspired by michael crichton, called Panopticon. Its set in our time frame and is all about encryption, three letter agencies, and a race against time! Here's a link with the first 20 pages if anyone is interested! https://docs.google.com/document/d/1VRI4X5fCUpwurUDvKmvzJpT7...
I am building the roots of an open source app for decentralized intelligence. Think git but for intelligence. Building it for myself since I'm becoming forgetful.
After years of having the New Year’s resolution, “make a personal site.” I finally got past my procrastination and did it! It’s live and I’m filling it out with things I’ve done over the years.
I am building a TCG card management app as a way to keep my machine learning skills sharp and to learn iOS development.
Pretty standard CRUD flows + plenty of opportunity to get comfy with iterating and deploying software on iOS.
This app will basically be the playground for me to work on:
- Frontend / UX
- Computer vision (scan in cards w/ Camera)
- recommender systems (what cards are good suggestions to add to this deck?)
- Clustering (what decks are similar to this deck?)
Websites like TappedOut and Archidekt are great products, but as someone who plays casually and doesn't do much to keep up with releases I find assembling fun Commander decks from singles to be quite a time investment, and I'd like to see if I can focus in on that UX specifically for myself.
A Jenkins replacement in Python (I hate Jenkins with the fire of a thousand suns, and it will never die as long as nothing appears to take its place). I've dropped it for the moment while I'm building a custom truck camper (which will also go on GitHub once my FreeCAD plans are done)
I just (like today) started working on an open-source, self-hosted, e2e encrypted, and free password manager.
I’m writing the backend in Go, with a standard API layer to allow custom frontends. I will also be building “official” frontends for mobile apps, desktop apps, and browser extensions.
The main reason I’m doing this is because I don’t like the UI of other self-hosted password managers, and I hate relying on the security of cloud options.
Seeing as I just started work on it today, I don’t have much (not even a real name). If you want to follow along, heres the Github, https://github.com/dickeyy/passwords
My goal for this project is to provide teams and individuals with the ability to secure their passwords while also providing a clean and elegant experience.
I had an idea for this you might want to consider adding: store the passwords/keys in the client, no server… but then shamir’s secret sharing to shard them along your contacts.
So when you open the app it asks you to choose five (?) friends you trust (who don’t know each other) from your contacts and it’ll ask them to install the app to help you store your secrets.
Then if you ever lose your phone or whatever when you reinstall the app you just have to remember at least three (?) of the five people you trusted and then will be able to restore your keys!
It would also ping the other contacts when open to make sure they still have your shards and redistribute them as you add/change/delete them.
I’m working on a tool called Together Gift It. Together Gift It makes managing group gift events like Christmas easier. I’m making it because my family was sending an insane amount of group texts during the holidays, and it was getting ridiculous. My favorite one was when someone included the gift recipient in the group text about what gift we were getting for the recipient.
Together Gift It solves the problem just the way you’d think: with AI.
Just kidding. It solves the problem by keeping everything in one place. No more group texts! You can have private or shared gift lists, and there are some AI features like gift idea collaboration and product search. But the AI stuff is still a work in progress.
It is a layer on top of the qtile window manager that allows for more advanced window arrangements - especially subtabs. It was an itch to scratch and I use it myself.
I am now diddling about with Rust and thinking of trying my hand at a PDF parser library. It's sort of nice when there's an established specs document - like all the requirements are clearly laid out and you don't have to spend time thinking about the scope and product design.
I can implement as much as I want to, have tests, and come back another time when I feel like it.
For the past few weeks I have been working on Enhance [0], a SaaS to help game developers easily add leaderboards, friendlists, store players' data, in-app currency store and purchases, and marketing tools such as email list and website builder and hosting.
I am almost done with the landing page and don't think I'll put much more into it, and will continue working on the API itself. But I still plan on adding a playable game into the 3D scene just for the sake of it and to learn. But so far I'm struggling at projecting a 3D camera render to a texture inside an already existing 3D context.
I’m a few days late to comment here but I’ve been creating metal furniture again this summer. Mostly tables, end tables, night stands, some benches and stools. It’s a hobby thing and my house is getting quite full, but I’ve also had a few commissioned pieces.
I like to mix different materials into the designs I cook up. For the tops I use cut stone which is prohibitively expensive for most hobby stuff but works really well for commission work because of all the different types (marble, granite, composite) and styles/ colors you can get. It gives it a very high end look and feel when combined with a sturdy metal frame. I’m also really into 3D printing so I try to incorporate that into the work as well. Using different pattern schemes to make inserts and accent pieces.
I’ve also experimented with incorporating IKEA furniture pieces into metal frames because I love the idea of doing modular furniture that can be swapped out for different colors and textures. I use the same idea for my 3D printed work.
I love the creativity and flexibility this gives me and my welding skills have ramped up significantly. It’s been years since I welded seriously and I really lost the techniques so getting that confidence back is great. I’ve been approached to do some big 5-6 figure home design jobs through the connections I’ve made acquiring stone pieces but I’ve only consulted on advice. I still feel a bit insecure about putting myself out there in structural work. I see how successful different design firms locally are building this stuff for tech companies and I’d love to eventually have my work featured like that but taking it slow for now.
Working on the concept of Computer 2.0 idea by Andrej Karpathy.
Basically it is an AI computer. It is nothing like we have seen before.
May open source it soon. With working zero hallucination (need additional evaluations).
Works with all devices. Created a MVP, air-gapped for military use, but it can be used for anyone including businesses/consumers who need an on-prem AI computer.
It has zero lag or zero second delay in responding your questions, and does not use any third party AI cloud platform. This is critical, because some of our customers have restrictive requirements for their data and they can't have anything of their customers to be placed on these cloud platforms. (eg. OpenAI, Google, Apple).
Adding additional features including Self-Teach, Cloned Personality, the holographic UI Wear, makes calls on behalf of the user, uses voice imprinting, edit videos, autonomously, for content creators and remembers 100 years of your life. And, more.
The UI Wear uses the exotic sensors I used in military to make it light and performant unlike Apple Vision Pro. Imagine interacting with something like "Jarvis" from Ironman. Quiet similiar but much more grounded in a sense that it's simpler and logical. Also, you can take it with you to play fast action sports or play games like Dragon-Ball-Z type game with your buddies. And it lasts a day, not two hours, on a full charge.
You can interact with the Personal AI Computer without the UI Wear, but it takes you to another level of experience.
Currently, doing it solo and looking for a cofounder who has experience in large distribution, logistics, creating polished commercial hardware, etc. Even if you don't these experiences, please contact me, if you want more info, interested, and/or want to work for this startup.
I’m working on Cyphernetes [1] which is a new query language for working with the Kubernetes API with a focus on highly connected operations. It’s inspired by Neo4j’s Cypher and views Kubernetes as a connected graph of resources.
It allows querying multiple resource kinds via their relationships (i.e. replicaset owns pod, service exposes deployment…) and easily crafting custom response payloads.
Lately I’ve introduced aggregation functions and the ability to visualize query results using ascii art.
This is a daily driver for me and I’m now mostly focused around features that will make it a complete kubectl alternative.
I've been passionate about superforecasting and AI for a long time, so decided to try and use LLMs to structure data about the world -- commodities, politics, etc. It's now a startup we're trying to get off the ground: https://emergingtrajectories.com/
Most of my days are spent reading the news and working on LLMs, which has been a blast. As an example, here's a dashboard that tracks major supply and demand shocks to various commodities around the world: https://emergingtrajectories.com/c/commodities
Thanks for asking! Yes, but we're actively addressing them.
We do a few things under the hood to make hallucinations significantly less likely. First, we make sure every single statement made by the LLM has a fact ID associated with it... Then we've fine-tuned "verification" LLMs that review all statements to make sure that assertions being made are backed up by facts, and that the facts are actually aligned with the assertion.
It's still possible for the LLM to hallucinate in this process, but the likelihood is much lower.
I'm working on a platform for recording and managing user acceptance tests for software projects. I am trying to build it out slowly but in the open at https://github.com/golang-malawi/qatarina
Besides building the ideas I have in mind around how UAT could be improved for both devs and clients (users). My other goal is to encourage more Malawians to consider Go as a programming language for shipping solutions in, my Go evangelism has been working slowly but I think people need to see something they can appreciate to get more people onboard.
Also working on a personal finance app in Flutter for young Africans, but that's not public yet :)
I'm still working on our privacy-friendly web analytics tool Pirsch Analytics (pirsch.io) like a madman :) I've been doing this for 3 1/2 years now and I'm still super motivated, especially because it's challenging from a technical point of view, and very rewarding (I live off it now).
There is still a lot to do and learn (especially in the marketing department), but we have plans for a new product in the privacy space. I don't want to say too much about it until we've started working on it, but it's in the compliance space and fits quite well with our existing product. I think it's always a good starting point to solve your own problems.
It's meant to be a scripting language to be embedded in another program, generating HTML. Eventually I hope to make a CMS and this be my solution for easy templating.
Technical details: Written in C, the language has an x86_64 JIT compiler, but falls back to a bytecode VM so it should work on any architecture. The language itself is dynamically typed and garbage collected. Currently at 20K LOC (not counting blank lines or comments), with good test coverage and checking for memory leaks. It's been quite enjoyable!
We are building a hybrid aircraft to connect cities with high speed door-to-door transportation, without train/airport infrastructure.
It’s a powered 55-seater auto gyro - an electric helicopter during take-off and landing and a conventional turboprop regional “plane” during flight.
We are at a “works on paper, let’s start prototyping” phase: basic assumptions are verified by people from industry, a development plan is in place.
My co-founder is an experienced high-power electric propulsion engineer and he’s currently developing our power train.
I am in charge of the business/operations/relationships side.
We are currently looking for a third co-founder to lead aircraft design.
When thinking about take-off/landing you have to take into account the powertrain, fuel, noise, power-to-weight ratio e.t.c. With modern electric motors, batteries, BMS’ e.t.c. we are at a point where system weight is comparable (the battery is relatively small, motors are quite light and powerful), but all the other things - noise, efficiency, time between overhauls e.t.c. is much-much better.
Long-term vision is to build-out a network connecting large metropolitan centers to smaller places, unlocking swaths of real estate. You can live in Albany and fly to Manhattan/Boston/Syracuse in 30 minutes whenever you need to report to office/go to a concert/visit a non-emergency doctor’s appointment. Vertiport can be located on a floating pier, on the roof of an existing transportation hub, on a plot adjacent to a bus/tram terminal e.t.c.
I've spent many hours under helicopter blades and can assure you that most of the sound comes from the blades, not the engines, when landing, hovering, and taking off. Also, there is going to be a point where you have to transition from an electric powertrain to a fuel powertrain and that handoff has to be absolutely perfect while hundreds of meters in the air. (Think grinding gears in a manual transmission car.) You could do something like a liquid clutch or something, but that is one more thing to fail and introduces inefficiencies. It is honestly probably better to go with traditional jet engine technology, at least at first, to verify the business model. Then, spend some time innovating.
I'm willing to bet that innovation in blade design will get you much further than innovation on the power train itself.
The beautiful part is that our powertrains are completely separate. You fly up/down using rotor. You fly forward using turboprop. The only transition that occurs is change in pitch of rotor blades to go from “powered” state into “passive autogyro” state. No complex clutches, no rotating mounts (like on Osprey). Completely separate independent systems. It also provides some measure of redundancy.
You are right that most of the noise comes from blades. And yes, there is some awesome blade design progress, especially for auto-gyros. Also there are a lot of specific things like low rotational speed, lack of tilt and other stuff that plays well into less noise in our specific application. But no, we won’t be as quiet as multicopters, especially those with ducted fans. That’s ok, this is a reasonable trade-off - manageable noise levels, a lot less complexity, lots of lift.
Not yet, we are a bit too early for that! But I sure hope that we’ll be able to start actually building soon, and then we’ll be very much doing that in public. It’s companies like Astranis, Boom and Hill Helicopters who inspired us and we’d love to follow their lead!
Thanks, I will look out for more updates here on HN.
This kind of vehicle has fascinated me for years. Seems like they'll be the transportation mode of the future, and that fully self-driving commuter aircraft will come before their ground-based counterparts, since the sky is a much simpler realm than the ground. The main challenge seems to be safety / reliability. Turbulence dampening, emergency landing in a variety of terrain, automatic ground-obstacle avoidance and communication with other aircraft (and that requires near-perfect cybersecurity).
We are obviously thinking about that. VTOL capability makes our aircraft a bit more safe - it can literally land anywhere but water. And there’s probably a way to make it float two.
But! Air is not an easier environment, it’s just different. With lot’s of limitation, both natural and regulation. My good friend is developing an automation suite for small aircraft and he’s dampened my optimism significantly! But still, It is the way to go, it’s just not as easy as it might seem.
https://matcha.video is an in-browser transcription-based video editor. Remove silence and keep only the takes that you want with just a couple clicks. Export your edits to an XML format that you can bring into FCP or Davinci Resolve to finish. Free to use, data stays on your machine.
Note: this is a WIP, so if you have any experience working with .fcpxml, I'd love to hear from you!
I love this. In a quick test, I noticed the transcription missed on quite a few of my "filler" words that I might want to edit out, although it did get some of them.
It took me a little while to figure out how to "edit out" words.
EDIT: Turns out should I have read the instructions before uploading my video it would have been straight forward :D
I appreciate you testing it! The whisper model that I'm using indeed doesn't seem to pick up filler words, I'll try to add those ASAP. I'd also like to add background noise removal as a preprocessing step in case the input audio isn't the cleanest.
Were you able to import the result into your editor?
Awesome! I think that makes you matcha's first official user. Please let me know if you have any other feature requests or use cases you'd like me to look into
Professionally, I recently founded a startup focusing on education in dentistry. I've developed a platform to discuss and interact with cases online which I use for my own teaching, and thought other might also be interested. I've made some of the custom components (image editor, 3D viewer) freely available:
https://kigun.org/
Personally, I'm building a website which lets you quickly setup personal online dashboards, because I wanted to have an overview of all the sites I refresh often on one page:
https://frankendash.com/
I'm currently working on a PDF search engine that will index and then search through several thousand pdfs. As a tabletop role-player and former researcher, I've always wanted a simple tool to do this, and I hope I won't be the only one to benefit from it. This is my first project of this scale and I hope to be able to show it here soon.
On the technical side, it's a desktop application coded entirely in java with a javaFX interface. No AI or online data uploads, just the Lucene search engine and PDFium for PDF rendering.
I'm working on Xona AI (https://xona.ai). I want to democratize the generation of 3D visualizations for interior and exterior design. We will launch next week on ProductHunt :)
Working on 8 inch Cassegrain telescope F 30-40 with small secondary and optimized for planetary observations.
But in order to get there, I need to rework a couple of mirror grinding machines in the club according to specs by the professional optician teaching us.
This means building the right set of mechanical thingies to safely hold glass and tools from 6 to 16-inch diameter optics. According to our optician, we need an aluminum disc the same size as the mirror with a reverse threaded hole in the center of the aluminum disc. This disc will be screwed onto a coupling which is attached to the shaft of the turntable motor.
We are currently working on sourcing the parts, machining, and doing a trial run with a 6-inch mirror. Then I get some blanks, get the curve I need generated to save me hours upon hours with 80 grit hogging out the glass before he can settle down and let the combination of grit, water, moving glass and time work their magic on the surface, every minute bringing me closer to optical eh-its-good-enough".
We have enough members in the club building telescopes that the effort of making these mirror grinding and polishing machines work is well worth it. With any luck, you'll see me at Stellafane next year.
I'm working on a small website for finding restaurants and breweries when you're traveling. It currently has 4 year of Great American Beer Fest winners and 3 years of James Beard nominees and winners.
It's very alpha at this point with no styling. I'm also working on getting addresses, websites, and instagram links gathered and added. Plus collapsing down the entries, so a place with multiple awards only shows up once, with all of its awards. Right now it's just 1 results per award.
I'm currently road-tripping through New England and it helped find the amazing bakery Norimoto.
I was going to point out the Samoan Cookhouse[0] - "the last cookhouse in the west!" but they've been closed for renovations for over a year. They'd been in pretty much continuous operation for over 115 years when they shut it down.
I hope they open back up, that place was a treasure.
I also did a road trip and had a lot of mapping ideas. I collected a few links on my own website at https://culi.page/maps/. The page asks you for your location and then plugs your lat/long into the url parameters of various useful mapping sites. Here's some of the maps I used
Ofc there's a lot more and many more I haven't added to the site yet, but hopefully you get the idea. I have a very large collection of maps/sites that can maybe possibly be useful or interesting and would like to have a single directory to easily access all of them from.
A database for my digital life. A place where to store semi-structured data and blobs, everything from phone numbers to my browser history, mail to photo and yes, also notes with a tagging system.
The primary feature that I'm implementing p2p synchronization (similar to syncthing, although the prototype use tailscale); append-only storage for easy backup and an "api" for easy integration/synchronization with external sources like RSS feed for example, all working offline.
Right now I'm using various software to manage my digital life (Zotero, Keepass, Syncthing) but i want to consolidate since I'm having trouble keeping all properly synced and backed)
I just aggregate everything on Git and found that it works so far. Although some listings (short stories, movies) are growing a bit too much for simple files. What's the lowest level database/filesystem you have to store your data?
Not sure if I did understood the question correctly: currently use a sqlite database as a backend for storing data, since it make easy to iterate and inspect while I'm developing, also sqlite seem to be able to handle blob just fine (up to 50MB).
I'm planning to have pluggable backends but the primary one will be a append-only log structure within a single file, however it should be relatively easy to add a git based backend. The reason for using a custom file format is that i want add support for a couple of feature to make it easier to recover data in case of corruption of both the data and the filesystem that host the file.
A small tool to align two versions of a video with each other, that have the same video (e.g same cut of a show or movie), but different audio tracks, so that the audio track from one can be merged into the other. The primary use case is to merge native BluRay rips that only have native language audio with dubbed audio from DVD or VHS rips. Should help simplify my media collection and cut down on storage.
Uses the scenedetect Python library to locate scene breaks and then uses perceptual hashing to find points for alignment. Also will include a small webapp to debug and adjust things manually in difficult cases.
I'm currently working on an open-source project called checkd, which is a Cloudflare service worker designed to handle device authentication and state management using Apple's DeviceCheck API. The project includes an example iOS app and a worker implementation to demonstrate how the service can be integrated into a mobile app environment.
This project has been an exciting way to explore better securing my iOS apps. I'm looking forward to refining it further and would love to hear any feedback or suggestions from the community!
I'm working on a news analysis project where I'll have to identify articles that speak about the same event. For now, I think it'll be based on date + "asking an LLM", but you probably came up with something better. How do you approach this problem?
https://elwood.studio -- Open Source platform for producing, hosting and monetizing video & audio podcasts through subscriptions, ads or one-time payments.
https://SharpAPI.com/ is a set of APIs to help leverage AI for workflow automation in E-Commerce, Marketing, Content Management, HR Tech, Travel, and other areas.
What can it do for its users?
- Product categorization for well-organized product catalogues.
- Generating engaging product descriptions and tagging.
- Filtering out spam content.
- Allowing understanding and analysis of sentiment in product reviews, comments, and users' Feedback for data-driven decision-making.
- Generating complex job descriptions and extracting information from resume files for easy processing - and more.
I've also created a set of SDK Clients for it like PHP, Laravel, .NET, Flutter to make it a plug-and-play solution for developers.
We're engineers, we want to make data-driven decisions about what we name our children. My app won't necessarily help you choose a name, per se, but it can assist in eliminating a lot of possible names, giving you a much smaller set of choose from, each of which you can research more. So... like filter by name origin, length, popularity, etc..
I was actually playing around those files today. I know this will be a typical HN comment and I'm sorry but, if it takes you an hour to load that data into sqlite, you're probably doing something very wrong.
$ sqlite3 x.db 'create table foo(a,b,c)'
$ time for year in {1880..2023}; do
sqlite3 x.db '.import yob'$year'.txt foo --csv';
done
real 0m1.359s
user 0m1.013s
sys 0m0.323s
$ sqlite3 x.db 'select count(*) from foo'
2117219
(Numbers are from Linux/tmpfs/i9-12900k)
Obviously that needs tweaked to store the year in a column, but it shouldn't run for more than a few seconds. Those csv files are the furthest thing from "big data".
Right on. Outta curiosity, whatcha up to with the name data?
Fair. I'm sure there are waaaay more efficient ways to load that into the db. That said, it's one of those things - like, where do I wanna spend my time -- sorting that out, something I only gotta do once a year and I can basically push "go" and walk away -- or something else. It's not like I gotta keep turning the crank.
I'm hunting for names that had resurgence in popularity because of TV shows/movies. The name that inspired the project is Mabel. It disappeared in the 1960s, then reappeared the year after Gravity Falls came out and it's still climbing. https://www.behindthename.com/name/mabel/top
It was built with WordPress, and the new theme is using Tailwind and AlpineJS. I hadn’t used AlpineJS before, so this project was a great way to experiment with it.
I don't have any monetization on it, and I'm not sure if I'll pursue that for this site.
Next up, I plan to build a Chrome extension. I have limited experience with that, so it should be a fun challenge.
Yes, I've been considering both options! WXT seems to fit my use case better at the moment, so I might give it a try.
As for stream names, I’ve looked into it as well, but from the limited research I’ve done, it seems like only paid services are available.
I doubt there's a way to get the radios' programming since these are mostly small radios with very basic websites and online presence. Best I can get is an MP3/AAC stream and that's about it.
I am building Clace https://github.com/claceio/clace, an app server for containerized applications. The goal is to build something like Nginx Unit, but supporting any language/framework. Each app runs in a separate container. App updates are done using a blue-green staged deployment.
Clace already supports most python based apps (wsgi or asgi), any other language works with a custom Dockerfile. Plan to add support for automatically shutting down idle containers, allowing for scaling down to zero for each app.
AnyStack Pages is aimed at making it easier and simpler to host static websites with free SSL protection and CDN. Websites can be hosted by uploading a zip file of the site contents or a single html file. It's currently in closed-beta for limited users every week.
Next Steps: Full launch with support for adding custom domains and premium features.
Happy to chat about this if you are interested in getting early access or for any questions and suggestions.
A text-yourself productivity app called tetr. I have always been messaging myself notes and thoughts on Signal and wanted something that would extend this habit to a funky productivity app and also do more advanced stuff like reminders, habits, mood tracking, etc.
Been working on this for nearly 2 years now and it’s been cool seeing people use it daily, some definitely a lot more than I do. I still feel like it’s not even close to its full potential; infinite more things to add but most often a joy for me to work on.
It's an online development platform similar to other things like Replit, but it's mainly for static site hosting. You can create projects and make custom subdomains for them. You can add libraries to your project in a click of a button, so you don't have to add scripts or CSS files to every single file of yours. I've recently been doing a lot of work on it and I hope to launch it pretty soon.
A VR game. Still in the concept phase though. It'll be free as it's only a hobby project. It might never materialise but it doesn't matter. Most of the reward is in the process and the learning opportunities.
For work I'm working on AI implementation. Because the vendor does all the technical goodness I'm kinda bored. It's also quite a crap solution IMO and we're kinda forced to be all positive about it because the vendor is closely watching and they have a lot of buyin from the top.
I set up a local LLM server too just to feel like I'm actually doing something.
Gnuplot is awesome. I've been hyping it up for years [1]. It's simple to get up and running with it, and remarkably deep when needed. Plays really well with poor man's command line data analytics (e.g. cut and awk).
Thanks for the link! One of the quite astounding things about gnuplot is that it continues to make steady progress, decade after decade, not stagnating or flaming out, without massive attention or lots of institutional support, and with great backwards compatibility. From the outside at least, it seems to be a very well run project, with admirable long term focus.
Then, I started working on an interactive TUI app for Python beginner exercises. I've previously written such apps for CLI tools (grep, sed, awk, coreutils, etc): https://github.com/learnbyexample/TUI-apps
https://subatic.com/ A video sharing site. Why? To save cost. Each video chunk is stored in Cloudflare R2 which reduces our streaming cost by 99% (no joke). Read about the calculation here: https://subatic.com/story.
https://readry.com is my latest venture ever since I got into e-readers/kindle. I want to read everything on my kindle and as it turns out, there are no good ways to get anything else apart from "books" on the kindle.
With this I can get my newsletters, top subreddit posts, news, and more, all on kindle. maybe this will help me curb my smartphone addiction.
I built a midi controller. Rather, I repurposed a Teensy-based virtual analogue synth I made ages back that never worked quite right. I had taken the amp/speaker out to use somewhere else, and it was just sitting in a box. I realized I could just use the Control Surface library (https://github.com/tttapa/Control-Surface) and change it to a midi controller in no time flat, so I did. Now just need to get some time to use it to play with VCV Rack.
I'm building a free library of website components and templates: https://uisual.com/. People can simply copy-and-paste the Tailwind CSS, Figma, and Framer into their own projects. No attribution, no sign up required.
I'm a designer, and this is my first Vue project. Super happy and proud of it. This is a way for me to learn coding. I'm working on improving the project by adding a simple website builder where people can edit the components and templates directly.
Loved this when it was shared in its own post recently. One piece of feedback is the bottom toolbar covers some of the content on mobile. I think ideally some extra padding on the bottom of the site would be needed when on mobile. Example: https://imgur.com/a/VXdtJ3m
I made a Chrome extension that *assists* you in doing Leetcode problems. It's like having a buddy give you small hints and ask questions to guide you to the best solution yourself instead of giving you the answer immediately! I've had many people tell me it's helped them a lot. https://chromewebstore.google.com/detail/leetcode-buddy/bled...
I'm working on a trip planner for vegans. This is designed to solve my particular problem as a full time (vegan) nomad who likes to stay in walkable cities.
I noticed that sometimes, I book accommodations in a new city without realizing I'm very far from the nearest vegan restaurant, having to walk for a really long time to get food, and sometimes getting there after the kitchen is already closed.
The planner helps me find hotels/airbnbs that are within 15 minute walk of at least 3 different vegan restaurants - makes my life so much easier!
For fun (and no profits!), I built an AI-powered haiku poetry and generative visual art app. It started as a daily puzzle (https://haikudle.ai, previously shown on HN), then evolved into an authoring tool with curated daily haikus (https://HaikuGenius.ai).
I just started working full time on my mortgage loan officer SaaS tool again. I had started building it when I worked for a mortgage company in 2017 because I was annoyed at having to enter the same data on 6 websites 20-30 times a week. Ended up getting some paying customers but somewhat unconsciously avoiding selling too many, a personality defect I have since seen.
If you know a mortgage loan officer, tell them to email tyler at lightningestimates dot com and mention HN. I'm just trying to fix an industry problem at a reasonable price.
A flutter based solo rpg that is a Hack of IronSworn. It is set in a world 40 years after an event called the arkfall, and where alien Arks crashed landed on earth bring with them many new flora and fauna and terraforming the earth at the sites of the crash.
The setting is like retrofuturistic version of 1998 with survivalist elements.
I’ve been enjoying flutter, and Have found it easier than other cross platform GUI technologies I’ve used. Dart is pretty easy to learn too and I’m considering it for other non flutter based tools as well
I'm getting to know the ins and outs of GitHub CoPilot, and exploring languages and technologies I've avoided. I'm hoping to get done with a usable Bitgrid Emulator[1] in Javascript/HTML so that I can let people play with the concept and get used to it. I've got stuff working in Pascal[2], but that's not something most people can deal with. I've also got a ton of other stuff up on GitHub that I should poke a bit.
I've spent a lot of time in analysis paralysis and this has given me the kick in the pants to get me going again.
As far as new ideas go, I've already spent time learning Verilog, and hope to get a chip design through the TinyTapeout[3] before too long.
Also, I found out that it only takes a few lines of Python and a lot of time to have an AI rate all my photos locally. That's a work in progress, should be less than 100 lines all told with niceties.[4]
I'm working on a CI/CD tool that runs scripts in the some of the most popular programming languages rather than YAML.
My motivation has been some recent jobs/roles in which no-pipeline-feature-gets-denied YAML scripts have grown unmanageably large.
I'd be interested to hear of any pitfalls that any of you might foresee. Other than this is a pretty hefty undertaking - that much I can already see on my own. :-)
About three months ago I had a moment of clarity when I realized that all I did with my spare time was work out, smoke weed, and play video games (not at the same time). So I decided to reallocate a portion of my video game time to learning artistic skills. Largely learning to draw with the classic book Drawing On The Right Side Of The Brain, but I've also been making music and video games and learning 3D modelling in Blender. I'm not good at any of it yet but I'm having a great time.
I’m working on Polynomial (https://polynomial.so), which is a small dashboard for key business metrics. It has open source integrations to most popular systems and has Google sheets export in case you want to do something with the data.
As a founder, I really needed a simple place to centralize all those business metrics. Couldn’t find anything that suited my needs (everything was way overkill) and so I ended up building Polynomial.
I'm working on a metadata catalog aimed towards scientific data. After standing up an enterprise data platform/lakehouse for science and laboratories, I've been really depressed by the current offerings and the massive oversight they have when dealing with scientific data. So I thought I'd change that. We're targeting not only the right kinds of data, but also connectors/integrations that reflect where most science data lives at these labs.
We were both tired of asking AI for the same code again and again, so we thought, why not build something where we can instantly save code from our AI chats
The idea is to create a public library of AI-generated code that anyone can search through, find what they need, and either use it directly or customize it further.
For example, if you find a function that almost fits your needs, you can add it to your AI chat with just a click and tweak it, instead of starting from scratch.
So far we have -
1. Chrome Extension - To save your code or find what you need from the library and add it to chat. Currently only works with claude, working on making it compatible with v0 and chatgpt
2. VS code extension to find whatever you need right in there and an AI that will understand your project's context and integrate these snippets in seconds ( AI part is in progress )
3. Website to do other stuff - build your profile, rate other snippets ( which helps us to maintain quality )
Making a fast, easy to use native work chat app @ noor.to
It's a way to share messages, files, code snippets, links, without worrying about distracting others, or quitting the app yourself to avoid notifications. It uses around 10x less RAM than Slack/Discord/Teams and prioritizes fast moving teams over enterprises. High quality software over features. Features that work rather than half-backed list of smart/AI/notes/wiki/etc bloat.
Working on visualising sailing performance by scooping data out of the onboard network (basically the same as the one in cars) and uploading it to a web app. https://www.see-sailing.com/
Looks interesting! Haven't done any sailing since leaving Auckland, but here in Barcelona there are also quite a few sailing yacht clubs. There might be some serious market for this. Good luck!
I've been working on an exercise database, using ChatGPT. I got a list of all the interesting exercise qualities, then I turned those into measurable data fields, and now I'm organizing the 600+ fields. The goal is to build the ultimate exercise database with 50K+ exercises.
But lately I took a break and worked on running (GPX run file analysis) and habit tracking. The eventual goal is to package it all together into one integrated PWA / mobile app.
I wrote a musical (or otherwise) pitch guessing game inspired by my partner's love for minimalist, but engaging, games like Wordle.
Still tweaking it, and it could use some better instructions. But the thread's here now so why not.
If you see confetti, you answered correctly. Otherwise the points are added or subtracted in a "hot or cold" fashion based on your answer's distance from the actual pitch.
I have started working on some open source actuarial tooling. Mostly relevant for non-life pricing.
The open tooling is quite lacking in the sense that one has to create a lot of stuff manually. Especially for the evaluations, even though most of them are quite common in the actuarial realm.
Also I wanted to tinker with Jax. So I am basing all on Jax, polars, seaborn/matplotlib and scipy.
I already have lift plots and Poisson regression in Jax. I’ll keep adding to it over the next months as time allows.
current pain points: the race starts immediately upon connecting. no time to prepare, so i always start in the last position because the AI drivers have a head start due to my slower reaction time.
the AI drivers are to aggressive to prevent overtaking, they will bump into the player, knocking them out of the position and causing a crash without any penalty to the AI car.
bumping into each other should cause both cars to crash, giving the player more time to recover. AI drivers should try to avoid contact however
Im working on MLJAR Studio, new Python editor perfect for beginners. It has set of interactive code recipes with Graphical Interface. You can create scripts with it. Additionally, it automatically install and import missing packages. It is a kind a new way for visual programming. Most of recupes are ML and AI focused. You can read more on https://mljar.com/docs
I'm trying to bring the encryption benefits of MTLS, the security of X509 Name Constraints, and the improvements that have been made in the various clients libraries and operating systems to TLS behavior into a nice group, so that any small company, or even individual, can run their own CA infrastructure that puts the bug enterprises to shame (specifically, https://enroll.visaca.com/ already deserves a lot of shame).
Building more Amazon scraper websites. Working on a framework to minimize the effort of creating them, and currently testing to see if I can automate most of the process with Anthropic’s Sonnet.
When I’m done, I’m probably going to make a blog post or two just talking through the automation I figured out so far. LLMs are amazing force multipliers.
Nope, just selenium and pyautogui. The priority is to fly under the radar, rather than efficiency; for that I needed lower-level control. There's a bunch of other stuff I did to improve the anonymity, but to download I just open selenium at the url, wait for it to load, then use pyautogui to type 'ctrl-s' and type in the save directory.
I am working on ByteChef(https://github.com/bytechefhq/bytechef). It is an open-source, low-code, extendable API integration and workflow automation platform. It automates daily routines that require interaction between independent business applications. ByteChef maintains automation definitions in an easy-to-understand workflow format.
I'm building an open source implementation of the Firebase RTDB server, that will be wire-compatible with Google's. This means that all the (already open sourced) SDKs will work pretty much as is, over REST or websockets. Security rules will also be fully supported. (The only meaningful caveat is that it's just the RTDB, so you need to deal with authentication yourself, and kinda shoehorn it into the existing APIs.)
Outside of that, I've been blogging a lot more (https://www.jacky.wtf/essays/ - August looks so full, ha) and now I'm writing about things I'm reading too (https://www.jacky.wtf/links/). Been doing this to try to ween off social media and rely on places like this to share stuff.
It comes with a set of tool for your IDE (currently only VSCode) where you can draw sprites, compose music, make SFX right next to your code.
It's still not in v1, it's also my first "serious" C++ project, and life got in the way lately -- I haven't made significant progress in the last month or so
I have been working on porting concepts of the Windows 10/11 taskbar, tray, and window management, to macos. With emphasis on simplicity, speed, and function inspired by Linux Mint MATE.
It has been a wild ride. Often frustrating but rewarding. Some days I may spend 10 hours solving an API for which app is frontmost at a given second. Sure, there's a system API for that, but it doesn't even remotely cover edge cases that exist. And many times with no answers out there on how to do something, it feels like I'm the first, so finally solving it is a great endorphin rush.
Part of the complexity is I wanted the app to be completely pluggable, kind of like Obsidian. The app communicates events to a local socket server in full duplex, which enables cross-plugin communication (but also cross-app!). Plugins access the app's JavaScript system bridge API for pubsub and system calls through a Webkit interface. Edits are hot reloaded and instant, no compile time necessary. The first time I could change my "Start" menu in real time through JavaScript/CSS was quite a feeling.
Sadly, I couldn't leverage existing tools like Tauri or Electron. They don't have adequate system bridge APIs available, and would be more work to leverage instead of less. They are too general, whereas this project only builds to macos. Therefore it can be much more complicated (and useful) by design.
I originally set out to be more productive in macos. But I've also spent time making prototypes for fun. A desktop widget system, real-time system color theme set from Spotify album art, live video desktop backgrounds from Twitch/YT, a Destiny 2 macos system theme, etc.
I plan to open source it and build a community around it one day.
To some degree OS productivity is subjective based on what you grew up with. Your neural pathways form around a certain way of work management and it becomes hard to change that after ten years.
I'm naturally biased then to early Windows and MATE-esque environments. And, it's worth noting that I despise Win11 overall, so a better comparison is indeed Linux Mint MATE[1]. Part of the project inspiration is to never use Windows 11 again, actually!
Before I started the project I wrote down what I consider productivity boosters for me:
1. Fast context switching between open apps and windows, natively. All open apps are in front of me by way of the taskbar, always, and never hidden. I never have to think about where to find my immediate work. I can group apps, pin them, and create custom behavior for them.
2. System tray apps for things that you interact with often, but aren't necessarily immediately working on. Macos has something similar but it isn't really pluggable or widely used yet, and nowhere near as customizable. With Win10 I can add, remove, or hide system tray apps based on how I use my workstation best.
3. Optimized Start Menu. I press a button, I get access to things I've favorited, recent apps/files, and categories of apps I use often. It is highly customizable in Win10, while I struggle to have something as efficient in macos (even though macos global search is wonderful!).
I feel like the old UX rule of "don't make me think" applies heavily here. The macos app dock is a great example of this. You're forced to think about what you want to do with 10-20+ options glaring at you. Disabling the dock is the first thing I do on a new Mac ;)
Lastly, Win10 lets me customize my system to a high degree as my needs change. I just don't get that with macos, nor have I found apps that hit the mark for me.
I'm curious about your own experiences if you care to share :-)
I'm working on a podcast and audio media monitoring and research platform. Easily the most complex thing I've built. But still so much more to do to make it truly useful. Testing PR and public affairs use cases at the moment.
I used to work for a media monitoring company but was instantly struck by how old fashion everything was. And the total reliance on boolean searches meant only experts could find relevant information. This still appears to be the case for most players in the industry.
So I'm building a platform that finds what is important before you look for it. Novel entity linking, and sentiment analysis plus speaker tracking. It has come a long way from the proof of concept. Focused on audio media at the moment as it is the hardest to index compared to news articles in my opinion. And the hypothesis is audio media such as podcasts can contain so many juicy insights.
Next steps are converting pilot customers to paying customers, testing more markets (based in a tiny market now), raise a small pre-seed (bootstrapped at the moment), and quickly evolve the product based on feedback.
yes, definitely some skin deep similarities! We both transcribe podcasts, but what we do after that is very different ;-) (also, i'm a fan of his podcast).
There is also more generally Mention, Brand24, Meltwater, etc, in the media monitoring space. But all are generally weak in the audio media space.
When we stop looking at audio media like a newspaper article, things get more interesting.
You can even buy the app itself on it. It’s going to be my starter for all other projects too, so I’m adding all the common CRUD app features. Also it aims to be lowest cost possible, which is free for now. I try not to tie it to specific technologies when possible to make it as easy as possible to switch providers in the future.
Read-only makes it so a buyer can’t push to main or modify the repo configuration. So to stop someone from paying to “grief” it in the worst case, and just to stop buyers from making changes the owner didn’t sign off on in more typical cases. They can still raise issues and submit PRs, to my knowledge.
Yes it’s intended for private repos. As it stands you could add a public repo but there wouldn’t be much point unless you’re using it as a way to accept donations, though I think GitHub Sponsors is more apt for that. Typically when you add someone to a private repo, they get full write access.
A database that basically acts like a modern day “virus scanner” for deep fakes. I believe that virus, malware and such that trick people into handing out information- making false decisions as we have seen with companies would be better suited if they had something like a heuristic threat score assigned to them. Yes we have websites that can tell us but it’s not the same as a virus or malware scanner. I think we need to add it to a database. Block those sites to protect consumers, companies so they don’t make bad choices. Just like in malware we see if a signature , deep fakes from bad actors also have signatures. It’s so many things. From signal- coding- so I was working on that. Maybe foolish or a waste of time. But I am going with to. Was playing with app development for cars to protect data. But might be a push against Nissan- Ford- Chevrolet-Kia- etc who like to steal data to make money off it. They control the interface. No VPN are allowed and no encryption allowed on infotainment system. So you have to do work around to keep your data out of their hands. That’s all.
It doesn't save anything to the cloud.
It supports markdown.
It currently doesn't using an LLM but I'm sure that will sneak in at some point.
One day, I'd like to have it generate wireframes using AI.
I'm testing out generative cartoons and media. I want to see what is possible with AI and media and if the content can be engaging. Long term I want to create content that could be used in a holodeck.
Here is what has been published so far: https://www.youtube.com/@distarkmedia
Would love feedback :)
Hey! I'm working on Getgud.io, an AI-powered game analytics and anti-cheat platform. Our goal is to provide complete observability into player behavior, detect cheaters and griefers, and help game developers improve player retention.
Some key features we're working on:
- AI-powered analysis of in-match player actions to detect anomalies
- Customizable rules engine for automated responses to toxic behavior
- Visual replay system for reviewing flagged matches
We support server-side integration for popular platforms like Unreal Engine and Unity. For integration guides and SDK references, visit our docs at https://github.com/getgud-io/getgud-docs.
Happy to chat more about game analytics and cheat detection if anyone's interested!
This looks rad, and interesting to see you have a lot of CS in the examples. I'm assuming this is because you get to just grab a ton of data from demos rather than trying to get integrations?
I'm trying to make it easier for Blind people to get into electronics as a hobby. There are many fascinating hardware, software, and UX challenges here that almost nobody is working on. Things like reading circuit diagrams, identifying pins on components, debugging circuits, laying out things on a breadboard, identifying wires, identifying resistors, testing components, and much more are largely unsolved Despite all the barriers, there are some Blind people who build electronics projects, sometimes including soldering, but they usually have to develop their own tools and workflows almost from scratch.
Since I'm just one guy with 100 ideas, I have to try to apply my skills where they can have the most impact. My current project is converting existing circuits from Fritzing format (which is usually displayed graphically) into a textual format. I wrote about it in these three blog posts:
I'm also working on a 3D printed single character Braille cell that can be added to any Arduino project, and has only 3 simple parts to assemble. Unfortunately I haven't posted about that yet.
I am building what I like to call PAI. Personal AI that ingests all your online information into a knowledge graph and is connected to your tools, all running locally. Will eventually build an off the shelf consumer device capable of running all this locally.
You can talk to it over a e2e encrypted reverse proxy (using Signals double ratchet protocol) so you can access it on the go safely and privately.
I have been working full-time on making staying at Airbnbs cool again with Hostup (https://www.hostup.ae)
We are doing this by providing 5-star amenities like massages, access to local experiences, events, and even essentials. Currently on the MVP stage, and only piloting in Dubai.
It's got long ways to go before being "complete", but I'm enjoying the heck out of working on this. I like working on things that aren't tied to money/serious job because they remind myself of the joy in programming :)
Mowing the lawn and cutting some branches without touching git.
I'm knee deep into coding 6 days per week as I'm building a startup and working right now on an "easy" solution for merging xls (yes xls) from some external company with our internal json, and I needed to let my keyboard rest for a day.
It's convincing me that our company should allocate a yearly budget to donate to OSS for stuff like the xlsx package.
Been working on a small temperature and humidity sensor using cheap components and a Raspberry Pi Zero 2 W. Using https://gokrazy.org/ to turn it in to a little appliance.
I've gotten quite carried away with a web interface and custom on-disk (on-sd card?) storage format based on Facebook's Gorilla paper.
I've realised that where it'll be deployed it will only have access to a public guest WiFi, so a remote server hosting the dashboard makes sense. So now it sends the compressed time-series over a TCP connection to a server component, hoping to bypass WiFi captive portals. Might need to use UDP to make it look like DNS or perhaps NTP. I haven't tested it on-site yet.
It has been a very fun and rewarding project so far. Looking forward to deploying it and getting remote updates working if I can get it to work with Tailscale on the guest WiFi. If anyone has any tips on circumventing captive portals and sending really small amounts of data through, I'd love to hear it!
Also, I have created the Vendor Stories - https://stories.ptah.sh - the curated list of success or failure stories when people are meeting the huge corporations.
- a fact management tool based on a timeline. I have a complex legal situation, so I created a tool to store information about events. I can prioritize and tag the events, attach files to them, and have small workflow on them, and then filter the timeline based on the tags/priorities/workflow states. It is very helpful, because the amount of events and data was too much to handle mentally. I am using Django/HTMX/AlpineJS.
- a dance event calendar for dance events in Finland, where both artists and venues can add events. The data is stuctured, which allows me to make views for each performer, venue etc. Mostly Django, but some HTMX/AlpineJS as well for complex screens
- for the dance event calendar, I have created a support site that uses ChatGPT to handle two first steps for incoming support emails/questions. First it categorizes the question, and then tries to answer it based on FAQ. If it fails, it summarizes the issue (and possibly asks few basic questions related to it) and forwards it to me.
Interesting. There was one app I liked but Bose bought and it seemingly went into the bit bucket. I like the person-led tours such as exist in SF but it's often morw convenient and practical to have on-demand computer ones.
It was Detour. My idea is continue where they left off by generating tours instead of having people produce them. I want these to be super personal. You might have a favorite movie that was filmed in the city you are visiting. I want to generate tours for that.
Okay this is crazy, but when I went to your website I looked at the church picture and thought "I recognize this!". I live at Forest and Waverley :D I will definitely going to add Palo Alto to my site and try and replicate this tour :)
Starting with cities and AI generated guides is a way I want to bootstrap this project. I strongly believe the best guides will be those created by people who are passionate. Your work here is proof of this :)
If you are local to Palo alto still I would love to chat with you about the idea. In any case thanks for sharing here :)
Not my website. A friend of mine from school who was also one of the lead Infocom game designers. But feel free to ping me. My email should be in my profile.
I'm actually a Cambridge/Bostonite though live about an hour West. Peel free to ping me about local stuff. Some local Boston movies (e.g. Good Will Hunting) are pretty weel documented, others are in areas I probably wouldn't go out of my way to visit, but lots of good routes.
Cool, thanks! I did check but don't see an email there. I just added my own contact to my about box. If you can send there or update profile let me know.
I quit my job last year as a CIO and have been working on my own consultancy focused on using strategic foresight to help develop better strategies.
I built a bunch of consultant led tools (Jupyter notebooks mostly) using LLMs to help with this, and saw how customers reacted to what they could do. So I decided a few months ago to bundle them together into a SaaS app.
I’m just about to release the beta of https://www.portage.so to my waitlist in the next couple of weeks.
The main use case is using a virtual whiteboard (ReactFlow) to string together a series of nodes that do a different task through the strategy development process. For example, you can generate scenarios, then analyse the impacts, then create courses of action in response to potential disruptions and so on.
Building LLM evaluation suites. Basically trying to test LLMs for privacy problems (data leakage/memorisation, PII extraction sort of thing), hallucination(RAG, summarisation etc) and security/compliance stuff(like bias/fiarness, toxicity, jailbreaks/prompt injection).
Involves a bunch of reading research papers, figuring out which ones are relevant to enterprise customers and getting our ML team to build it out. The most interesting part of this is how you present the insights of a given test to a customer in a consumable and usable format. (Ex: Just dumping a bunch of RAG hallucination metrics isn't enough but you want to figure out what are the key insights and interpretations of these metrics which could be useful to a data scientist or ML engineer)
Vertical tabs for firefox and lots of features. I use it myself in all my firefox profiles. It can be displayed as a sidebar or a popup. Which means you can use it in big windows and small windows.
Trying to expand on a quirky travel website populated by AI content. Working on adding countries, regions, areas (e.g. ski resorts), etc.: https://meoweler.com
The quality of the generated content is surprisingly good, but there are many stereotypes and mistakes that I'm fixing in the next mass generation pass.
I'm working on writing a sort of code and data playground from scratch - with currently only CodeMirror, ProseMirror, and some icons as dependencies - so that everything in it can be edited live. Everything is web components with the shadow DOM. It is sandboxed both as a whole and at the notebook level. The overall sandboxing uses a Content Security Policy that allows inspecting the URL for following a link, to enable using untrusted code (from humans or an AI) and private data without it easily being leaked. I've built a resizable split view, draggable tabs with multiple panes, and a color picker. It's a few dozen Markdown files with code in fenced code blocks and a Deno and Docker-based build process. There is a hybrid notebook/playground editor with the fenced code blocks changed into links that open the files in tabs. It's open source and at https://ristretto.codeberg.page/
Hi, I am Software Engineer from India, I am working on https://shoutmemo.com to collect, manage and showcase customers testimonials using embed widget on website. It can also import testimonials from socials like twitter. This is my first side hustle. It is still in beta stage.
Thank you for asking! Excitingly, I'm working on maintaining and improving my canvas library. Which is not news because I spend much of my spare time doing this exact same work.
However for this months work, I'm adding new stuff to the library - specifically some new OKLAB/OKLCH filters: something I don't think other canvas libraries have yet got around to considering[1]. And this weekend (a 3 day weekend in the UK) I'm spending time fixing the horror-show code for the reduce-palette (dither) filter[2] to see if I can get it to work just a little bit faster ...
(Fans of small PRs should definitely not click on these links)
this example seems to simple. if i write that, why would i not write the invitation myself?
ok, i get it, it lets me avoid having to address the person directly, especially for people who are insecure or just overthink what to write, but is this really healthy?
where will it end? using AI to write love letters?
on the other hand, you could leave out the email sending feature (and this avoid any spam problems) and just send the generated text as a reply that i can then forward.
a generic email interface to send AI prompts sounds useful. encouraging people to write personal emails using AI not so much.
Building Mujo, a minimalist musicians' practice journal for iOS. It integrates time tracking, note taking, and task management into a familiar metronome UI/UX. I dearly love it (1400+ hours logged!) and it's slowly building traction. https://mujo.app
Hey everyone! We have a struggling economy here in South Africa, so working on an interesting way to help consumers save money by onboarding product pricing and helping people track their savings while still choosing their preferred brands from stores nearby that are cheaper.
I'm using this as a project to learn Swift and SwiftUI for iOS, then I'll get my team to rebuild in Flutter if the prototype yields interest.
The idea is to focus on a small community, and give back to that community.
The working name is Savery. :)
If this is successful, it could be launched in other places, in which case we'll likely open-source the code and have communities manage their own systems.
My context is that I an a founder in a solutions company in South Africa, and I am learning to build things as a hobby to be able to best relate and discuss technical things with my software engineers.
https://hypv.ai ... prototype ecom platform like Big Commerce etc. but with all the Algolia-type predictive stuff baked into the bare metal so can target content at a much more granular level than trad platforms and has visibilty on much higher-resolution signal data
I have been working on a little price tracker for the Hetzner Dedicated Server auction (https://radar.iodev.org/). Using Sveltekit for the first time, and decided to go frontend only with duckdb-wasm.
I pull the data every hour with a GitHub action and redeploy the site nightly. The duckdb is regenerated every time and I can flexibly change its structure as required. So far I didn’t change much (eg. no normalization) since it’s not really needed. It also turns out to be surprisingly small, with compression a few megabytes. I haven’t written much SQL in some time and duckdb is very powerful.
Most of the texts are AI generated as I’m usually very bad coming up with this. I really gotta learn that one day. Webdesign is the other weakness, but with all these UI libraries these days it’s less of a problem.
I guess it’s pretty niche, but maybe interesting for some.
Hey all! I'm working on a context-aware CMS called trilly.
The idea is fairly simple: the client app requests content and passes the context (key-value pairs) along with it. In the trilly app, the content manager sets up different variants of the content for various context field combinations, and the one that matches is returned. Context can be anything, from user role or user location to more abstract things like A/B group or user behavior. It's just data.
I have a couple of ideas on how to extend this system - realtime LLM-generated variants, or more sophisticated context fields (date/number ranges, randomness, etc.).
For now, however, I'm exploring whether this idea would be useful for someone, so thanks for all the comments.
Outside of working hours I’m developing a podcast hosting platform called Soapbox[0]. I had started my own podcast with my college friends and rolled my own RSS feed and site with AstroJS[1]. Then realized I could offer the service to others. Thanks to modern web APIs I’m going to pursue an in-browser recording studio with WebRTC and then in-browser basic audio editing. I’m very excited because it’s my first side project I’m trying to turn into a business (I opened an LLC and everything). I’m gratified because it’s very different work than my day job in fintech. I’m also offering internships to my college’s CS students and mentoring through that as well which is very fulfilling (I’m a recent grad).
I’m working on an Open Source SQLite VFS extension with modern encryption using libsodium. Using the VFS instead of adapting the source like SQLcipher is not straightforward but feasible. I hope that a modern encryption library, extensive testing and using vanilla SQLite source can convince users to give the extension a chance.
I've been working on Habitat for the last 4 months. It's a self-hosted social platform for local communities. The plan is for it to be federated, but that's a while off yet.
With ImageTool, you can:
- Search multiple stock photo sites with one tool
- Generate 500 images with AI
- Edit images
- Compress and convert images to .webp
It also helps you bulk edit images.
The goal is to help web developers speeding up their workflow.
I just published the first issue our digital zine, Forest Friends. The first issue is on "LLM System Evals in the Wild".
Lots of AI engineerers are doing vibes-based engineering, just eyeballing the LLM output and saying "LGTM!". This is a good place to start, as we all should look at our data more. But it's best to move on from vibes to system evals.
The first issue is on how to design and build system evals for a systematic way to gauge how well your LLM app is doing. That way, no matter if there are new models, new users, or new queries, you can be sure you're continuously improving, rather than allowing regressions.
An app for easing house management - HomeHero (https://HomeHero.pro). Primarily inventory oriented towards easier finds, location hierarchy and related tasks.
Currently only web, android and Linux as there's been little interest from Apple and Windows side
I'm working on my tech interview platform: https://onlineinterview.io/
It is a free platform that allows people to develop on the browser using multiple programming languages, draw on a shared board and talk over video.
I'm working on https://runs-on.com, a self-hosted solution to get 10x cheaper GitHub Actions runners with support for Linux x64 and arm64. This month I've been refining Windows support and hopefully this gets fully shipped before Aug 31st!
We're really looking forward to Windows support - I don't think any Actions runners vendors supports Windows at the moment and we were looking at building our own runners as a result, but if this launches soon, we'd be very keen to try it out!
Working on plump.ai – job-seeker's AI hiring agent. Think you have a personal career manager who searches for better job opportunities, applies you to them, speaks with recruiters on your behalf up until the interview scheduling, and does it 24/7 without involving you until real human action is required.
The ARC-AGI challenge / Arc Prize. It helps me enjoy programming even after becoming PM. I got a bit rusty on structural induction for trees, it’s now all fresh again.
After playing with it a bit I am now convinced a A* on the right kind of AST structure will be sufficient to make a significant breakthrough in the scores
This weekend, just designing a new type of lithography exposure process platform for hobbyists. Getting sub-micron 3-axis repeatability is an interesting problem, as one explicitly considers where to source economical mechanical precision.
Mostly just bored waiting for my Tungsten sample shipment to try in a new metal 3D printing process (at $90/in^3 you figure people would just air mail it, but nope cause it is a fire hazard.) Container rail transport work stoppage means I may get time to entertain low priority vanity projects.
Also working on a physics inertia puzzle so embarrassingly stupid... I refuse to post it lest it wastes 1 minute of someones time.
Kind of like this ear worm... these kind of problems are annoying... lol =3
I’ve been working on a project to make SaaS (and really any software) pricing and packaging easier. Think things like plans, entitlements, dimensions, metering, subscription migrations, onboarding, and experimentation, all completely independent from billing and payment processing solutions. We can integrate with things like Stripe where we’ll mirror products, customers, and subscriptions, but it’s totally optional.
We’re 100% bootstrapped by way of a previous acquisition and just very-soft launched after talking to a lot of folks at companies of all sizes. Seems like most people end up cobbling this stuff together with various levels of sophistication, which is basically what we did a few times over at previous companies with varying levels of success.
Recently after working as a frontend engineer for more than 2 years, I started doing full-stack, so I am going through the language that my new org uses (they are using GoLang) and trying to understand backend concepts as well.
My first task was to write a GitHub action to build and push the docker image on AWS ECR. While working on it, I went through a good number of blogs, and also used ChatGPT, and finally raised a PR. So I thought to write a blog on that, you can read it here - https://hsnice16.medium.com/build-and-push-the-docker-image-...
https://joinsymbol.com — Manipulate text with translations, JSONs, and other revisions with AI. Provides an API for using in your apps, etc. CMS on top AI basically.
My current strategy is to go inch deep, mile wide. I don’t want all of my eggs in one basket any more. Basically doing a VC model using my portfolio of products.
Holler at me if you want to chat about these products or if you have any ideas you want to build.
I am trying rebuild my virtual assistance business. For the longest time I have worked with realtors but the market has gone down and finding new contracts are hard.
Trying to move into doing virtual assistance for micropreneurs and sideprojects from now on exclusively. Task by task basis or regular contracts. Essentially the pitch is assign everything that is not sales and product. I think it is a good pitch targetted to folks who enjoy building stuff and considers making money out of it as a bonus. The moment they start answering support emails, and posting product updates they start to get burnt out. Trying to specialize in "burn out prevention" tasks mainly.
I am trying to figure out a way to get clients. Virtual Assistance is cheap and the competition is huge. The services are identical which essentially says, I will do what you will need me to do.
Hello! This year I've been working on building an integrated registration and live timing platform [0] for my local organizers in CA, but it is spreading. Very fun/rewarding project. I hope to launch my own cellular-based timing hardware this year which will integrate into the timing platform (right now I've reverse engineered and now support 5 different chronometers, I never knew cross platform serial port comms was such a PITA). Getting cellular based timers synchronized should be fun, although I'll probably opt for some super simple solution, we'll see...
Also continuing to build FastComments [1] and hoping to continue to grow that.
Looking for feedback -- we had some good initial traction on HN, and are looking for OS users/contributors/people who are building complimentary tooling!
One of my hobbies is trying to predict foreign exchange prices (usually EURUSD tbh, but others as well).
I enjoy finding out the actual best-case performance of various technical analysis indicators, and popular "systems", by optimising the indicator's parameters to find the best long term performance. I also design my own indicators and systems in the same way.
To this end I have a C++ ImGUI desktop app with charting capabilities, and optimisation using differential evolution. The app has its operation driven by a c++ class that controls all aspects of its running, by means of various callbacks.
This enables me to test lots of strategies quickly, exploring the parameter space with the optimisation functions.
I've considered open sourcing the app, though I'm not sure there would be any interest. The codebase would need cleaning up first anyway! :-)
I'm working on https://currentkey.com/ - a way to assign names to your Mac's desktop Spaces. I've also been working on a couple of games in Godot, but they are pretty far off from being interesting-enough to share.
We are fanfiction and web fiction lovers. We built a Goodreads-style service that alerts you when your favorite online novels get updated. You can also create lists and review your favorite works.
Essentially paying down (with time and money) maintenance debt on house and property after semi-retiring. I'll get back into doing some "real" work in the fall. Seems to be a pretty common pattern.
I got lazy during COVID but stuff seemed hard to get done and I wasn't really in the mood to try harder.
I'm in a similar boat but from different reason. Stared working on HomeHero (https://homehero.pro). It's mainly focused on taking care of home and it's inventory so... maybe it will be useful?
Thanks. For me it's mostly about getting stuff out of the house and doing some necessary repairs/maintenance. I can mostly keep track of stuff. I "just" need to get necessary jobs done and I've really gotten behind.
After the recent NCAA settlement, college athletic departments will (in the next year) be able to distribute as much as $22M+ to student athletes — aka revenue sharing
We’re building the tech that General Managers need to manage their roster, valuate players, construct contracts, and pay players.
I've been working on Lorcast (https://lorcast.com/) a Scryfall like search for Lorcana. Scryfall offers a powerful search engine and query language to search for Magic: The Gathering cards. Lorcana is a new card game by Disney that also has a competitive scene. As the number of cards grows it will become more important to have good search _and_ good APIs for the community to build with.
It's been a breath of fresh air to work on something that I have no plans to monetize. I just want to build something useful for the community to use and that has allowed me to be relaxed with building it.
Anyways, if you play Lorcana I encourage you to check it out! The game is lots of fun.
I'm working on converting a jukebox to use a raspberry pi. I want to keep using the numberpad to pick the tracks and none of the existing solutions that i've found seem to do that. I don't have any experience with python and very little with linux so it has been an experience.
Keep going! This sounds like a really fun project that will be really satisfying when you start to get things working. Not to mention the experience gained will be valuable for sure!
1. I'm writing a book on Git. I did my research and was shocked to discover many otherwise tech savvy individuals still cannot manipulate Git from the command line, and resort to the use of desktop applications (yikes!)
2. An edtech app to learn coding efficiently. AI will certainly enable 100x developers, but we must first train 1x devs.
Features: structured learning, a curriculum designed by humans, AI assistance for when stdents get stuck, code samples and project based learning, covers different languages (we start with Ruby, Rust and JS, with more to be added), and technologies (CL, Nushell, SQL - again more to be added), numerous exercises. Think of it as an improved version of w3schools and the like.
One reason I am disclosing this online is to hold myself accountable.The other is in hope of finding an angel investor.
I'm developing a tool using GPT-4o to help draft outlines for small-business and institutional grant applications.
I'm also working on a solution to fix prior authorizations in the insurance industry, fix scheduling surgeries due to insurance network database inconsistencies.
I'm building a new kind of knowledge management app called https://saner.ai/ :0
Where you can connect all your knowledge sources, query them with AI and create new content based on them with many AI model
I am also thinking and planning about similar system for interactive and intuitive prompt based queries ala ChatGPT but it is for comprehension rather than generating new documents or data. The main aim is to have easy access to knowledge and information on various government documents on laws, regulations, procedures, etc, and since it utilizes some confidential documents the contents should not leave local HPC.
I just publicly released the first version of my
web+email+database combined server package that
I have been using myself for a while now for all
the different server based projects I have been
working on.
I try to keep it as small and simple as possible
while still being flexible and containing everything
in one package that you need to write server applications.
It also has no dependencies except for very common
stuff like a standard C library and a Unix-like
operating system.
A zero-dependency application in Bun. It has a tailwind-like layer which can be seen here [0]. It’s not production-ready, but has been an interesting experiment.
The Tailwind bit isn’t Bun specific, but the unit tests are, and the app currently uses (or eventually will use) various things such as bun’s http layer, password hashing, SQLite API, etc.
I’ve been realizing that while I’ve found an incredibly effective solution to a problem I keep having, it does not automatically lend itself to becoming a wildly thriving business. If the problem that is being solved is a somewhat infrequent one (once a year or less), and not applicable to a super-wide audience (think Facebook), then it simply won’t result in enough business to commit to fully.
I built a domain hunter (with availability checks) for .com domains based on idea prompts and while I see people purchase domains all the time through it, they won’t even shell out $3 to get any extras or a list of more than three domain ideas or so. Zero willingness to pay anything, but high success rates on search and result conversion.
>they won’t even shell out $3 to get any extras or a list of more than three domain ideas or so.
Are the extras worth $3?
I think this may be an interesting article...too cheap to be of value.
Of course, without knowing more, it's hard to judge: more info needed.
Simple and ultra low cost (sometimes free) loyalty program software for small business owners using Google and Apple Wallet. All the tech hassle taken away from them.
Tailored to local markets with language support, formats and habits.
For the client side FB, telegram or whatsapp sign up, no password, no nothing. Just get the card via qrcode and that's it.
Comprised of a web app for the SMB's and their admin staff (basic tenant setup).
Mobile app for the line staff (Android).
Also adding staff check ins to automatically calculate monthy hours.
All this is based on previous work with real coffee shop business owners.
So giving it a try as a SAAS service.
I made a personal, self-hostable read-it-later service that fits my needs. Just for fun I gave myself the constraint that I couldn’t use any JS, so it’s all done with static html templates in Django. It also archives all articles so you never have to worry about link rot. https://github.com/brendanv/lynx
And now, because I have a problem, I’ve completely pivoted and I’m rewriting it from scratch as a SPA because I wanted to try using Pocketbase for the backend and extend it so I could learn Go. https://github.com/brendanv/lynx-v2
A head-hunting/recruitment application to search companies and people based on their skills and experience. Currently focused on Australian market.
It's also for people who are looking for jobs. The goal there is to help to find good companies which have high (or low) concentrations of people with certain skills/attributes.
A no-code/low-code platform for building apps without code. Insnare is built with and runs on Saasufy. This freed me up from having to write back end code for Insnare and made the front end code very small.
I'm cooking... Essentially a way to mock up any layout with design for any interface (web, mobile, desktop, whatever) and then save the layout in a portable format so anything can read it. The data is formatted so that you can re-create the layout natively.
-Uses WASM under the hood (blazingly fast)
-Exportable into a simple format for desktop, mobile or web use
-
@htmx_org
integration included
-Pluggable into any existing front-end or back-end (React or Laravel, etc...)
For now, the work of re-creating the layout lies on programming but I will start building plugins so that it lessens or removes that need.
Just last week I started hacking on a k8s operator for managing postgres roles, grants and databases. I really like the continuous reconciliation for this scenario, especially having previously worked at places that managed this with shell scripts or terraform.
Nice thread! A lot of interesting projects. Fingers crossed for all of them
On my side, I'm working on https://flatcal.com that aims at simplifying sharing multiple calendars from different sources as one ical link. Like combining work, freelance, and personal calendars to display my busy time for easier scheduling etc. I'm finishing handling time zones which are messy as each provider has its own approach, but manageable. Expect it soon :) A have a huge list of usages for it so not short on ideas right now. But first things first, I need to focus on releasing the basics now. I'm having way to much fun with this one
I rewrote https://github.com/laktak/chkbit in Go because Python packaging was so frustrating. With Go you can just build for any platform without spinning up multiple VMs.
It is a budgeting tool that allows you to create budgets and visualize them as interactive Sankey diagrams. You can configure smart "pockets", which can automatically send excess cash or take missing cash from other pockets. It also allows you to invite other users to shared budgets, so you can use it to plan budgets with roommates or your partner. Currently I am thinking about ways to make it into a viable business, but it is still not quite there yet where it needs to be for that.
It's very nice and clean and functional. Budgets are some of the most boring and universal tech ever, and I think your project is very worth continuing. Kudos for having a collaborative (if not fancy and needlessly realtime) feature from the start!
I'm building FitBee (https://apple.co/4aGUw5X) - a simple and user friendly alternative to MyFitnessPal. It's been really fun trying to figure out how to build the most frictionless food tracking app possible. I just launched a feature where you can get a calorie & macro estimate by taking a photo of your meal and am working on several improvements to it.
I have a small amount of experience building iOS apps but have never done anything with a backend before. This project is fairly backend heavy, so it's been quite a journey learning how to do everything from setting up a database to deciphering elasticsearch.
It's basically like ShaderToy but for WebGPU instead of WebGL. I started it as I have been doing some Rust + wgpu development for art projects and I need a easy way to play around with shaders.
It's very early in development - you can go and just use it right now. But soon I want to support creating an account, saving / sharing shaders, and eventually go beyond the featureset of ShaderToy by allowing for custom input images / textures.
my list of features needed for a usable audio player:
for audiobooks the ability to skip forward and backwards a few seconds. the best way to do that i found is to swipe left and right. short swipe is a short skip, swipe across the whole screen is maybe a minute. also the ability to quickly jump to a certain manually chosen timestamp.
also for audiobooks, the ability to bookmark positions. not all audio books have chapter markers.
for audiobooks and podcasts: list files by their names in storage sorted by timestamp or name. many podcasts and chapterized audiobooks don't set chapter or episode titles correctly. i have seen players where the title was always the title of the book or the name of the podcast, making it impossible to identify which file was which episode or chapter. also author names and even podcast titles are sometimes inconsistent, because spelling errors, different authors or multiple names for a podcast.
do not assume that the file metadata is consistent or correct and make it easy for the user to work around inconsistencies.
Working on an NFL stats & visualisation project, the working title is American Football Insights.
It's a hobby project, I'm not planning to charge for. I did a bit of work on this a few years ago, and I found that analysing the stats, you can gain insights, which you can't really from just watching the games. It was prompted by the upcoming NFL season.
Things I'm thinking about...I had an idea for a centralised local DB where I collected all kinds of stats on my life, all in one place. They're already being collected on various platforms, I thought 'why not make them my own, and maybe there's something to gain by pulling them together?'.
We built a radar-based device for (gamified & fun) agility testing and training with accurate measurements, for top athletes. Works without wearables or expensive and complex/location-restricted setups like you have with timegates or LPS sytems.
If anyone here is into sports science & tech, feel free to get in touch! (laurens@ the domain above) We're looking for a sports scientist / product manager to strengthen the team, a marketeer with affinity for sportstech, resellers & partnerships, and of course, more customers :-)
https://synthql.dev: An alternative to PostgREST focused on the React-Node-PG stack. Main advantage right now is it gives you end to end type safety: you can write type-safe queries straight from your component that directly access your database.
Why SynthQL? My experience working for +10 years on enterprise SaaS is that a quite often you just want a database and a way to fetch data from it. Backends will quite often get in the way adding abstractions and layers upon layers of transformation between DB objects to domain objects to DTOs.
If you ever feel like you just want to talk to the database directly, give SynthQL a try :)
I've been working on a Hacker News TUI reader written in Rust [0].
I'm very proud of the way the architecture turned out - with most notably a components-driven architecture [1].
There's just a major performance roadblock in posts with many comments that I should be able to clear with some more multithreading. Then I just need to make it available in brew and other distribution solutions.
For like the past 2 years I've been building out a cloud storage solution that makes use of end-to-end encryption, but it tries to make organising and sharing fun, in that files are shared by tagging them with stickers. https://stickerdocs.com.
Working on my lisp interpreter and freestanding programming system. I've more or less finished my ELF patcher and am working on the loader/linker that will form the basis of the foreign function interface of the language. This also laid down the ground work for working with binary structures from within lisp.
I'm also studying electronics and trying to learn circuit and PCB design. I want to buy an uconsole and make boards for it. I contacted the uconsole's power management IC manufacturer and they actually sent me some GPLv2 Linux drivers! I intend to clean it up and port the features to the existing kernel driver so I can upstream it.
Removing duplicate annotations from the Kindle clippings.txt file.
I like to annotate (highlight and create notes) when I read on my Kindle, and then I take those annotations and throw them into Obsidian. The issue is that if you ever mess up a highlight or note and then delete it, it doesn't actually get removed from the Kindle clippings file. So this project will remove all of the "incorrect" clippings.
I have the de-duplication algorithm working extremely well and right now I'm working on creating the web app using python/django
http://cleannews.fyi - Just started this as a side project a few days ago but the goal is to get AI to generate a proper newspaper while striving to eliminate editorial bias
Thanks I had this idea for a while but only with the new gpt models I think its executable now. Is there a market for this? Either way fun project and I will generate a newspaper in 90s style everyday
latest success seems to be probiotic for oxolate-digestion, which now helps me digest green salad. For a test I tried green onions, but got anxiety back. Being able to digest onions and garlic would enrich my food choices a lot.
I doubt there's money to be made from all this material, except I could offer some coaching, but pretty much everything I know is in this little repo
Reading the repository readme, on one hand I feel incredibly sorry for you, on the other super excited that you've found something that works for you. Depression is such a wide catch all term and there are countlessly many underlying reasons that can lead to it. Unfortunately that also means that curing depression is always a bespoke effort, every person suffering with it is different so it's very hard to try and figure out what helps. I had no idea food could be linked so strongly! I hope what you've found is sustainable and leads you to a better tomorrow.
Myself! I ended up with an incredibly severe health issue due to repeated mismanagement/misdiagnosis by doctors over the last 5+ years, and now that I finally know what's wrong I can treat it effectively and maybe even recover.
I've been struggling to find a workflow that can easily extract knowledge and insights from audio content on the web and sync it with note-taking systems such as Notion, Readwise or Obsidian, so I decided to create a system that transcribes the audio, summarizes it and shares it with other applications.
Right now we are only targetting podcasts to Notion as the vertical slice for the MVP, but in the future we're looking to support "connectors" that can take in other forms of content such as audiobooks, videos, etc. and share it to other popular note-taking forms.
It's been an exciting journey so far and we're looking to launch soon!
Working on a new PaaS called Molnett (https://molnett.com/). Trying to replicate the development productivity you might have at larger tech companies like Google, Spotify and more.
Kind of struggling to wrap my mind around how we can avoid having EBS-style network disks as part of our product offering. We are convinced that it's a operational nightmare to maintain and it's better to provide other tools to achieve the same results. For example, instead of running Postgres on EBS or local disks, we use Neon to run it on top of S3!
I started publicly inventorying and documenting my collection of vintage print advertisements (https://adretro.com).
The front end CMS is a combination of Notion and Super.so. The backend will be Lucee and MySQL.
I have many thousands of ads I want to catalog just to start. There are so many print ads, cataloging just a fraction will take a lifetime.
I'm developing a custom cataloging system that will allow me to process them more quickly and automate aspects of the management of the collection. I have in mind to eventually make the database available to other collectors.
I'm current building web browser AI agents (can do anything the browser can do via prompting, logging in, scraping, can code websites with the data it scrapes etc.) going to ship next week! if anyone wants to join our beta lmk.
I'm working on a server driven UI framework for mobile apps.
Designers can push changes in Figma similar to a merge request, which can then be approved by product managers / developers and the apps UI updates on every device.
Devs don't have to touch UI anymore but can focus on pure logic.
Any app developed with this doesn't have to go through the usual release cycle for UI changes anymore.
This will decrease the time lag of mobile to web and makes it easier for us to implement and distribute campaigns.
We're in Alpha and two mobile apps with our technology are going to go live next months. A credit card app and a competitive player versus player mental arithmetic game.
https://zimly.app is an open-source S3 media backup app that I've been working on in my free time.
The last few months I've been trying to get the real-time upload speed to display nicely, which proved to be harder than anticipated. But I think I've got a good solution now and might roll a new release today.
My goal is to have a few thousands downloads by the end of the year. Marketing is hard, also because it's less rewarding than programming features, so I tend to pick a new feature instead of writing a blog post or polishing the website.
I have a few projects laying around. But currently I am working on an iptv player for Android. I learned about Jetpack and found it fascinating than the xml based old process. So I started playing around and learning Android development. And lo and behold, I ended up with a more than mvp IPTv player. Now I am considering if I should be putting it as open source software, or maybe publish it to the Google store. I am considering a Google Tv counterpart where you can share data between instances on the same network. That's one of a huge pain point to input data on the tv.
Working on an update to `hiedb-mode`. It's an emacs package for querying an hiedb index; a database for building IDE tooling for Haskell. The update will use `compile` for the display buffer similar to how `find-grep` does it.
Looking into better compression algorithms inspired by ML, LLM techniques under different constraints. For example if the receiver has a prior knowledge because future devices will have some foundational models that already understand higher level concepts like how a human looks or speaks, do we really need to send the entire recording or just the "words", "intonation" etc and reconstruct it on device.
Fabrice Bellard and other has done some work in this domain
https://bellard.org/
Working on a stock market strategy tester and portfolio simulator. Basically what a lazy dev would want to invest in stocks. https://www.equitieslab.com
https://article2audio.app/ is my creation. It's a BMW of article readers -- it handles images, tables, preformatted text, and many other nuances of a Web page. I built it to scratch my own itch. I invite you to a free trial.
This app can be useful for listeners who want to enjoy articles while driving, in a gym, etc., and for website owners who want to offer an audio version of their articles.
I just reprioritzed my remaining 300 or so tickets for my chess variant AI sandbox game https://www.chesscraft.ca
I recently released a major update involving presigned s3 urls, and typically after a big update I pick my favorite issues (always fun) for the next big update while I go bug hunting. Sometimes I feel like this project is code therapy where, unlike at work, I get to do everything the right way and/or my way.
I've also just started strength training at a gym for the first time which is already helping enormously with my neck and back pain.
I'm integrating daily steps walked into my workout/nutrition tracking app EverBeat for Android. I've noticed that this is an important metric for many people interested in losing some weight.
I estimate burned calories on the minute-to-minute level for all activities, which is a fun challenge.
For the steps to calories conversion, I require time and distance walked. Unfortunately it's surprisingly difficult to get some kind of distance information from Android. Google Fit API provides this data, but will be killed soon. The RecordingAPI will hopefully provide this in the future, but not yet...
I’m working on an iphone app to fight my screen time addiction. You pick the apps you want to block, set a timer for how long you want to stay focused for, and lastly you set a pledge.
A pledge is the price you’ll pay for breaking your pledge. It’s more than one second of reflection. You have to ask yourself “do i want to spend a $1 to use YouTube when i should be studying”?
I don’t have a website right now. I’m calling it FocusPledge. It’s near complete so if you’re curious you can shoot me a message or follow me on X at @harveenatwal_
I'm working on a framework to convert code into video, with a focus on helping creators and software educators. Ultimate goal is that you could automate nearly any educational coding video you could think of. Stretch goal is to fine tune an AI so that a step by step video could be generated from a blog post, book chapter, etc.
I've been messing around with an e-ink Spectra 6 full colour display. Colours are so much better than Gallery, but the contrast isn't as good. Still makes for better looking photos. I'm merging support in my DIY smart picture frame project this weekend (https://github.com/DDoS/Cadre).
I'm trying to get my hands on a larger display, but man does the e-ink public documentation suck. Still looking for some actual complete documentation and sample driver schematics for the 13" and larger models.
I've been building some macOS apps lately, and I kept duplicating and adding onto the license key backend. I figured I should build it more properly so I built a Saas for myself (and finally took the step into Laravel at the same time).
I have no costumers except myself. If anyone wants a free account in exchange for feedback I am happy to help set it up.
Goodreads is Amazon so its ratings are suspect, like everything else on Amazon.
It would make a ton of sense to have a third party rating and review site with some kind of dependable reputation for posters. You could still affiliate link to books for site revenue.
- Micro SaaS https://yasl.at, link shortening with custom meta data for link previews on messengers/social media, see my blog post for more details: https://yasl.at/dKpNnZ
- A web game called Wizencraft (https://wizencraft.com) which is a logic puzzle to guess the right order of items as quickly as possible which lets you compete with other players, solve daily challenges in groups and let’s you grind thousands of items, it's a very early alpha and I'm building it in public very iteratively and not polished at first, see more details here: https://yasl.at/XnoDtY
- And last but not least I started to build a not yet published simple dashboard for analytics where I can plugin my self-hosted database very easily. You can just write SQL queries for MongoDB, Clickhouse, Postgresql in a single YAML which defines the query and how a chart looks like on the dashboard.
The idea came from the fact that using other dashboard/analytic tools always is a hassle, existing analytics dashboards I used for products are not easy to copy, replicate or adjust and it's mostly doable solely via clicking in a complex frontend instead of defining it simply via a configuration and the available offerings are charging large.
I'm planning to have the chart definition editable via web interface and is stored locally as files which can be versioned and managed by git and can connected to the web interface with a simple docker container (this brings the analytics dashboard to the product source code itself and can be easily copied/adjusted again for new products). Happy to share more and once the first version of the dashboard is ready to use (in the coming weeks...).
---
Would love your thoughts on all of this and especially the analytics dashboard. Did you you face similar issues/problems with existing solutions (or if i can stop and actually use an existing solution which does what I'd like to have).
I’m working on a macos virtual microphone app that lets you take a real microphone input and apply audio processing effects on it like raising the gain, reverb, etc, then using that as your microphone in other apps, like for video calls
Comes from my own desire for something like that. Right now I’m using a hacked together solution using blackhole and a random vst. It was a pain to set up initially, trying to make it easier for other people.
I know there’s loopback but it costs too much for what i need and has a lot of extra features i don’t care about, plus i’d still need to bring my own vst to it
We've been building https://www.datafragment.com, a tool that can help you get prospects for your product. It lets you search the web's HTML code.
Some of our current users use it to find Wordpress websites with specific versions of Wordpress, others to find websites that use niche-market tools, ones that are not covered by incumbents such as Wappalyzer or BuiltWith. We're currently focused on the French market.
Curious to hear your thoughts, if you are looking for new ways to find prospects.
I work with my small team on the exclusive co-op game Sixth Force, about father and son. Father is AI network developer and son-teenager is the artist. They find themselves trapped inside an AI network simulation and need to get out before the network takes over the world :).
Btw your wish listing on Steam is very much welcome to finish the game https://store.steampowered.com/app/2830090/Sixth_Force/ :)
I'm working on a side project which is a Python library that provides of an abstraction layer to manage infrastructure in cloud providers by interacting with their APIs. This is something I've been developing every now and then for the last months after years of working with Terraform and getting tired of the limitations of it's DSL (yeah, Pullumi is better but I just simply want a generic library with classes representing services in the cloud I can call natively in Python without having to deal with a 3rd party application like Pullumi is!).
A companion that you chat with to keep track of your health symptoms. I was inspired by seeing family members go to their doctors appointments and not knowing the important symptoms to report in their short appointments.
Right now, it's Ruth, a companion you can WhatsApp with. She will take your history and record your symptoms, then you can ask her to summarise it before your appointment.
Try out the prototype on WhatsApp: +1 (516) 734-6593
(It's not HIPAA compliant and shares data with LLM providers currently, so best to use a fake patient profile if you're interested)
I found a way to add code to NFTs. The simplest way to understand it is that it is similar to Adobe flash.
I know that NFTs are down, but I have invented a way to add actual utility. I call them xNFTS - executable non fungible tokens. Imagine the famous Ape or Penguin NFTs - by adding my technology, the NFTs can change images, play music, or play games without installing anything.
I got a patent on the process and they work on any blockchain- or even off. the market is down, but I am very excited about tying web2 with web3.
If anyone wants to join or learn more, let me know
A form builder and filler SDK for developers. We make it easy for developers to build form experiences into your SaaS infrastructure. From simple documents and PDFs to large complex dynamic forms. It gives you the tools to provide the very best form solutions that empower your users on every level.
I'm working on https://mockoon.com an API mocking tool. I launched it here as a side-project 7 years ago, and I'm now full-time, trying to bootstrap a Saas platform.
I'm seeing some progress (I officially have some precious MRR!) but it's one hell of an emotional roller coaster...
I am trying to build a algotrading system for myself.
I have reached the stage where it acts as my co-pilot and I can read graphs and take trade, but I am now stuck to make it fully automated.
Hey HN! I'm still working on my Docker Swarm dashboard, Lunni: https://lunni.dev/
Lunni essentially takes docker-compose.yml and deploys it on a server. Currently we're using Portainer as the backend, which in turn just runs `docker stack deploy`. I'm reimplementing the stack deployment code from scratch which will allow us to support latest Compose spec and extend it as we need (and perhaps support orchestrators other than Swarm in the future).
I'm building a keyboard-driven editor for designing UI. You could think of it as a weird bastard child of Vim and Webflow. I've been tinkering with it for a while in private, but I'd like to shift gears and start working on it publicly.
The editor itself isn't available yet, but I've got a Github readme that explains the concept a bit more: https://github.com/matry/editor
Eventually I'll do a ShowHN, once I get a stable version that I feel comfortable demoing.
Minimal blog platform. I'm creating one for myself but I'm making it generic enough that other people could fork it and make one for themselves. PHP and vanilla JS, as usual.
I've been working on BatchWizard, a CLI tool for managing OpenAI batch processing jobs. It lets you easily upload files, create batch jobs, check status, and download results - all from the command line. Handles multiple jobs concurrently with async processing. I built it to simplify working with OpenAI's batch APIs.
Check it out on GitHub if you're interested: https://github.com/cmakafui/batchwizard
I'm working on getting long text summarisation to be useful, you can read some of the 50+ page AI research paper summaries here that are machine generated and are getting close to human generated ones.
Super excited to share one of my side projects that’s finally ready for action! Meet PowerTiger—a powerful, open-source energy monitoring solution built around the RPi Pico W and a custom built PCB. It’s perfect for tracking power consumption in real-time across 16 power terminals. Installed it at home to collect power consumption metrics using Grafana and Prometheus running on RPi home server.
I’ve been thinking/slowly building a service that hosts Kubernetes control planes. Bring your own worker nodes. Users can get a fully managed control plane (upgrades, HA, etc.) in their region of choice and can use whatever workers they want—be it cloud VMs, bare metal or your laptop too.
I’ll eventually open source the single binary agent that’ll bootstrap a host into a K8s node. Just run it rootless with a join URL and voilà!
Also in the pipeline is a global load balancer service (for your clusters).
Kubernetes is really built on the assumption that workers are in the same LAN as the control plane. Long latency between the control plane and workers affects workload reliability; heartbeats need to be configured with longer timeouts, for example. Pod-to-pod communication, where pods run in regions on opposite sides of the globe, supposedly on the same pod network CIDR, is also going to be flaky. There's a long history of projects attempting to take LAN-local designs and make them resistant to regional failure by superimposing a LAN on top of a WAN and it never works as intended. Furthermore, various service meshes already present ways of helping to direct/shape traffic between clusters (i.e. between regions), when service architecture evolves to truly support multiple regions.
You run a risk of building something that seems to "work" and falls apart for non-obvious, head-scratching reasons for many or most users.
> Kubernetes is really built on the assumption that workers are in the same LAN as the control plane. Long latency between the control plane and workers affects workload reliability; heartbeats need to be configured with longer timeouts, for example.
Yep, and that's why I'm designing this to provision control planes as close to the user's workloads as possible (sub ~100ms latency, which is plenty acceptable for reliability; the likes of Scaleway do similar).
> Pod-to-pod communication, where pods run in regions on opposite sides of the globe, supposedly on the same pod network CIDR, is also going to be flaky
Certainly, but this is not something that I as control plane provider care about -- it's a design decision the user has to account for, and I consider this freedom of choice a nice one to have.
I suppose the main sell here (and the few folks I've spoken to seem to agree) is the flexibility a decoupled control plane offers. Being able to migrate your bare metal setup to "managed" K8s; running a homelab cluster without dealing with the ops (upgrades, cert rotations, etc.); or simply being able to use VMs from cheaper cloud providers.
But yeah, it's definitely a hard problem but solvable within the right constraints.
I'm currently working on Merlinn, an open-source AI agent that helps on-call developers troubleshoot production incidents.
https://github.com/merlinn-co/merlinn
Web app development [1]. I know I know, it's 2024 and my friends are doing very cool things with LLMs, but I started a bit late in life and am just catching up with the 20th century (and I feel just fine about it :).
Just finished a blink reminder app https://blinktracker.app a couple of days ago.
Monitors your eyes' blink frequency with a camera and also reminds you to take regular screen breaks.
Somewhat ironically after giving up smoking I take too few breaks and get blurry/dry eyes after long days. The app is a bit overengineered and comes with an api in case you ever wanted to start controlling stuff with blinks. Currently Windows only.
Lately I have been working on making MicroPython an excellent platform for deploying Machine Learning on microcontroller-based sensors. Primarily this is via this open source project:
https://github.com/emlearn/emlearn-micropython
But there is also a bit of work needed in the wider ecosystem, like improving drivers for accelerometer/microphone/etc, support for data formats such as Numpy .npy files etc.
I have my first Swift/SwiftUI app up on TestFlight! I've always been slightly irked by how Google Maps, Apple Maps, and Yelp manage bookmarking places, so I'm creating a to-do list app that pulls from MapKit and automatically organizes your saved places by neighborhood and city for easy retrieval and in-the-moment decision-making.
Experiments with speech compression/processing. The modern approach to this (and pretty much everything else) is to transform the input data into a high-dimensional space, throw it into a neural net and cross your fingers. (the 'Whisper' STT runs in 512-dimensional space!). It works, but it's hardly elegant or compute-efficient. There's a certain art in doing things the 'old way', even if a million years of evolution decided neural-nets were the one true path.
it's a free CSV Viewer with charts that supports viewing data in both table and chart formats. It includes features like sorting, searching, filtering, and pagination in table view.
I like leaderboards, and its a bit sad that Redis is the only optimal database for leaderboards. So, I'm attempting a drunken version of a statistic database that tracks stats and leaderboards in realtime. I guess the only major defining quality would be that in one step I can get both the sorted scores + entire row of linked user data.
For redis this is two steps: `ZRANGE leaderb 1 100` + `GET x1 x2 x3 x4 ... x100`.
It's mostly a refresher for deep C++ doodling, but maybe I'll use it in production for kicks.
The backend is a web socket server written in go and the frontend is solidjs. Two technologies I've been wanting to use properly in a project for a while. Would recommend this to anyone wanting to improve their understanding of go's concurrency model. I've really enjoyed working on it, if anyone can think of any improvements please let me know!
Work in regulated manufacturing industry, deploying off-the-shelf operational technology systems (think Siemens, Emerson, Rockwell, Honeywell, Aveva, etc.).
Been building a software stack to automate building the custom documentation / validation the businesses who buy and deploy that software. Excited about its potential to cut down and the least favorite part of the job that engineers and those businesses have, when trying to improve things by deploying software or making changes
Smart assistant for kids that takes on the personality of their toys. Christmas gift for my nieces and nephews, so I am building four of them.
Raspberry Pi5 + RFID reader, touchscreen, and some 3d printed enclosures. Going to get a lightweight LLM to run on the Pi5 and some custom interactive software.
I just started this project about two weeks ago, but I've been tinkering with the hardware side of electronics lately and this is the culmination of about 18 months of various hobby projects.
I have been working on a platform for digital content creators where they can sell their work or build online communities. In a few weeks it will be exactly two years since I started. I had a 6 month break but beside that it was daily grind from morning to evening. I'll be going online later this year.
Most funny thing is that the toughest part related to the project had nothing to do with coding or tech. But finding a free domain. I had no idea 99% of the .com domains are just squatters.
I've got the API ready, which requires 2 main values:
- The original file ( which can be sent as a file, or hosted internally and requested using an ID )
- The data that needs to be printed into the audio file.
The API will return a watermarked version of the audio file that you can use later to extract the same data you sent before.
It's currently being tested on a production website, will wait for feedback, improve, and create an actual service out of the API.
Reading chapter 3 of taocp (in vol 2). Randomness sparks something in my brain for some reason. It's nice to see Knuth already said something about what is a random sequence.
I am writing about Graph Reasoning on my Substack (Encyclopedia Autonomica - All things Autonomous Agents) because I really find the topic interesting and there is not much research.
I'm working on QA Sphere https://qasphere.com/ - a lightweight tool for managing software testing. It’s built for QA engineers to document functional test cases and track execution, manually or via automation.
Test management tools are nothing new, but I’m focused on making this one fast and easy to use, without the bloat. The goal is to help smaller teams adopt solid QA practices without the hassle.
Right now there is an enormous surge in applicant demand.
Talent teams face the challenging task of handling high-volume inbound channels, while maintaining an excellent candidate experience, and dedicating time to high-value recruiting activities.
So I'm working on Hirevire(https://hirevire.com/) to help automate the screening rounds so that talent teams can spend time interviewing only the best candidates.
I've been thinking a lot about how to have healthier relationships with our phones. There are already dumbphones, but they're quite extreme and not everyone is willing to switch. So I wonder if there's anything else that could be done in between. Investigating and writing on my blog (https://nullderef.com) has been super fun so far.
I am working on https://hacktrack.info a SaaS that alerts you when a new CVE or exploit is published for the software you use. The idea came to me while working on an incident response team, where I noticed that many companies were hacked due to using software versions affected by recently published CVEs or exploits. Most of the similar solution I know of are really expensive or part of a larger product suite.
I am working on an app for creating stories with illustrations in a revealjs slide deck.
The idea is that a user would handwrite or type the first sentence of a story they want to write. This sentence would be used by llama 3.1 as a prompt to make a short story. Then the story is fed to stable diffusion to create image illustrations. Finally, all of it is combined into a slide deck that users can see in the app and also download if they want.
Chillin - Next-Gen AI Video & Motion Editor, I'm working on the online video editor, https://chillin.online.
Chillin has a significant advantage over its competitors(veed.io, clipchamp, capcut online), eg it supports mobile devices, offers full keyframe support, no watermark high-resolution video exports, and supports vector motions.
So I feel like I am changing the world through this work.
Hi guys,
For the past few months, I'm working on a form builder called minform with focus on UX and themes. Current form builders too restricted when comes to styling. It's free for unlimited submissions with paid plan for team feature.
We have been working on AI agents for sales and marketing. Our focus is in deep customer research instead of at scale marketing or spray and pray motion of outbound emails.
Our purpose is modeled to give sales and marketing skills to non-sales people. Just like how Canva gave design powers to non-design people.
We believe everyone has to have the skill to sell and market themselves or their product. Yet the only trend that does not change is understanding your customer/prospect.
I am currently working on building a lending platform. My goal is to both run it as a product loaning money to businesses and sell the software as a service to other investors as well. Very early phase, talking to a few friends who are currently in this line of work. I have a couple decades of experience myself but as an engineer and leader, not as a product or business owner, so if you have advice or ideas, please respond here, I would love to hear from you.
Working on making my first personal website since I was about 16; not totally sure how to get the balance of personal and professional down fully but I feel like it's going to be very good at encouraging me to refine my side project experiments instead of just dumping them on github and maybe posting them here.
Will probably try to self-host because why not.
Also doing my first Java based project since school; surprised how okay it's been going so far.
* No need to invent a new language, use Python!
* Use JIT to compile a SynthDef to a proces function!
* Eliminate block_size at runtime by compiling a version of each primitive for every 2^n block size.
* Watch the compiler do unrolling, vectorisation and merging loops.
I'm slightly pivoting https://uxwizz.com, to focus more on the text-to-SQL feature to generate graphs/charts and export them. It's a lot easier to create custom dashboards like this, and it's also easier for me to implement the "UI" to create the embeddable charts using AI text-to-SQL instead of a visual SQL editor.
I have been recently thinking about the NP-complete class of problems particularly the knapsack problem (it is easy to understand and attack). And I think it is really solvable in polynomial time. I actually tried and posted my solution here.
https://news.ycombinator.com/item?id=41341608
a bit of feedback would be really helpful.
Your solution is not an exact solution to the knapsack problem, but a (relatively well known) greedy heuristic approach towards the problem. It is a good approach towards a solution, but the knapsack problem isn’t as trivial as it seems.
For a counterexample, imagine you have a ton of items that have a weight of 10^-20 and value of 1, and a single item that has the weight of whatever is the capacity of the knapsack (so it’s only thing that fits if it’s inserted to the knapsack) and the value of +infinity.
All items have their cost ≈ 0, so whether you fit all of the “light but valueless” items or the single “heavy but extremely valuable” item depends on how you sort when costs are equal.
An improved heuristic would take the value into account more than just having it as a factor, but it’s not easy to design.
I would recommend finding an open course on algorithm design and analysis and going through the lessons while trying to extrapolate approaches to problems instead of just solutions to them.
> Your solution is not an exact solution to the knapsack problem, but a (relatively well known) greedy heuristic approach towards the problem.
can point me to where to find more about this.
> For a counterexample, imagine you have a ton of items that have a weight of 10^-20 and value of 1, and a single item that has the weight of whatever is the capacity of the knapsack (so it’s only thing that fits if it’s inserted to the knapsack) and the value of +infinity.
that is rather an extreme example. i do not see the how it affects the validity of the solution in most cases.
> I would recommend finding an open course on algorithm design and analysis and going through the lessons while trying to extrapolate approaches to problems instead of just solutions to them.
implementing solutions is the concrete result of extrapolating approaches and choosing the best one to implement. and that is the pragmatic approach.
Sorry, I’m not too familiar with English-language literature on algorithm design. I just remember this from my uni course on it :)
I believe we used Introduction to Algorithms by Cormen et al. as the basis, so it should be covered in depth there - but I’m not too sure.
> that is rather an extreme example. i do not see the how it affects the validity of the solution in most cases.
This is a fair point, but also a distinguishing factor between a heuristic and exact approach to algorithm solutions to problems. Algorithmic solutions usually have theorem proof-level rigor for exactness. Also, I just used the easiest counterexample to come up with :)
I am working alone on MyApps [1], an app of apps where you can spawn any number of pre-defined "app windows" into a zoomable, pannable infinite 2D space. Currently it only features a clone of PureRef.
It runs Gundb in the backend, providing a privacy friendly, p2p syncing organization tools.
I'm tring to write a minimal tool in Go to deploy and manage CD for my self-hosted projects.
I've tried many things but still none fit exactly my needs (I deploy many static websites so I want that to be effortless but I still want to be able to do continuous deployment on git+docker based projects). I know about Dokku and is probably what I should have learned to use but this way I can also challenge myself on a medium Go project
The RCL configuration language, https://rcl-lang.org / https://github.com/ruuda/rcl. Lately I'm working on support for floats to complete the json superset promise. Also I added a shorthand for "query --format=raw" to make it more useful as a jq replacement.
Working on verifiable correctness for programs written in LM or anything that generates annotated assembly. Basically low-level proofs that accessed memory is valid and live or that function pre/post-conditions are met.
The goal is that these proofs are compiler agnostic, so more people can use them.
I'm working on a project to find the best places for you to live (in the US).
Right now, people are deciding where to move based on gut feels, spreadsheets, and wikipedia pages. There needs to be a better way to enter personal preferences and come out with enough data that ensures the place you're in is the best fit for you.
Enter your preferences, narrow down locations, and compare them:
For my startup, a way to debug code errors with LLMs: https://codecomet.io - please let us know if you're interested and we can show you a demo.
On the side (with very little spare time), a set of Scrabble-related apps: for studying, AI, and playing. (See https://github.com/domino14 for more info)
Jobchef.io is a SaaS I've worked on for a little while and recently released.
I have a background in marketing and HR in large company, this is my first venture into programming after learning to code on my own (VBA -> Automate the boring stuff -> coursera -> React & a lot of help debugging from a very generous friend.)
Releasing a full working product was a great milestone, but so far market fit is still quite unclear.
I've been learning io_uring for fun and 0 profit by building a web-based game using only the Rust standard library. I create a simple WebSocket server and am now creating async from the ground up. I'm documenting it all: https://github.com/kilroyjones/series_game_from_scratch
Working on https://github.com/shivam276/smolmigrate, a simple migration tool for postgres databases, I have never been much into ORMs and was having hard luck finding migration tools tied to them which were also lightweight and easy to use, wrote my one..
Would be interested to learn more about how and why this was coded in Common Lisp, in particular which value it provides specific to the problem being solved compared to other languages.
Ability to write at a very high level (and macros), incredibly fast when compiled, ability to use the repl on a live server for diagnosis and patching functions, language design choices work extremely well for parallel recursive hierarchical inference.
Not working. The other day I uninstalled Skype from my cell because some group chat notifications had gotten past tolerable.
August is a sanctuary. It's the only thing I can hold every day of the rest of the year against Tech's meaninglessness. It's when I can read and do what I want instead of what I must.
I am working on an open-source self hosted alternative to railway.app, render, vercel. I know there are several similar platforms as alternative to these providers but the platform I am working on will be purely running on Kubernetes clusters, no ssh to VPS and use of docker-compose or docker-swarm.
I'm currently building a mobile-friendly Postgres database tool. Think dbeaver but on Android.
Writing and editing SQL queries on the phone can get tedious. Rather than just giving users a tiny textfield, I am exploring if there are better ways to build SQL queries on a touchscreen
I'm working on a Tournament Manager for Pickleball at http://pickleballtourneypro.pages.dev because the current incumbent is very difficult for non-technical people (aka volunteers) to use, and super expensive (10-15% of tournament revenue).
Lot of fun ideas, and hopefully will get used by a large variety of volunteers.
It's an incident management platform, similar to Pagerduty, Rootly or FireHydrant.
It's the first side project I've open sourced, and I've been hacking on it weekends and nights. Hoping to get a few companies to start using it to get some early feedback.
We're currently developing https://www.videototextai.com/ – ChatGPT for video and audio. The idea is to get to an all-in-one video and audio editing/insights platform. We’re actively building out new features to fully realise our vision, and we'd love to get any feedback from HackerNews!
I've been working on a better spreadsheet for a while now. https://rowzero.io is 1000x faster spreadsheet than Excel/Google Sheets. It looks and feels like those products but can open multi GB data sets, supports Python natively, and can also connect directly to Snowflake/Databricks/Redshift.
Just recently released version 4 of our free library for data deduplicating/linkage at scale. After four years working on it, finally getting to the stage where I'm really happy with both the performance and the API design.
I don't know if you are aware of Pijul. It claims to have solved some issues with git merge. It is one of the tools i want to study, use and test extensively at some point.
No, but from a quick glance it looks like an alternative to git? There might be value in studying their approach. But any solution I’ll be working on will have to fully work/integrate with a regular git workflow.
Oh, if operating within git boundaries is a requirement, then a different version control system is no good.
The claim is, that pijul's merge of code is much better and simpler than git. Merges are idempotent and associative if i remember correctly. It cannot get much better than that.
The reality is that most people are using git at the moment. I am looking to create a product that I'll be able to market and sell (rather something that I have fun R&Ding), so the git boundary is a requirement.
Working on my modern take on an API management + monetization platform . I want to make it super easy to sell API access with all the bells and whistles you need such as; authorization, rate-limiting, subscriptions, credit billing, usage based billing and a self-service customer portal
Well I am not a fan of RapidAPI for a couple reasons, firstly them taking such a big cut of an API provider's business and secondly their marketplace feels kinda spammy and low quality.
I am trying to build so that there is little vendor lock-in (you can export your users to other auth providers, billing is integrated directly into your stripe account, etc) so that you still have the choice to migrate if you ever feel unhappy with the service
More specifically, my mental health. I'm a bit of a mess, and I'm not sure I'll be able to properly commit to any external endeavors until I'm feeling better inside.
Kudos to you! Even just being more aware of your mental state is a step in the right direction. Reminds me I need to be more proactive in this area.
Dunno if it helps, but just know most of us are at least a bit of a mess. Life is messy. I once took two years off of regular work (we were fortunate to have enough savings to do it) and made two music albums, the first album based exclusively on hard things that either I or people very close to me went through. The process of fleshing out the songs and lyrics was incredibly therapeutic for me.
> Dunno if it helps, but just know most of us are at least a bit of a mess.
Years ago, I tried to quit my job because I was feeling terrible about how I was doing. My boss gave me one of the best pieces of advice anyone's ever given me:
"You know, Rachel, the suicide rate among founders is..."
This probably sounds like one of the most unhelpful things someone could say in that situation, but it was EXACTLY what I needed to hear. I didn't need someone to say "oh, it's ok to feel bad". I needed someone to say "feeling bad doesn't mean you're not strong enough to be great". I needed to hear that ambitious people who are trying with all their might to be better are still constantly struggling with the idea that they aren't good enough.
YC's own founder school doesn't quite put it in those terms, but there's a LOT in there about the importance of human factors. Arguably more than there is business advice, actually. They explicitly say, for example, "don't pick a co-founder who has complementary skills, pick one that won't drive you insane, because breakups and not lack of skill are a more common failure mode". And there's like half a dozen videos on how to keep it together when you're stressed beyond breaking. It's pretty reassuring.
To really make the point: my company became profitable for the first time yesterday afternoon, and I spent a good chunk of the 24 hours since then ruminating on how now we might fail because I'm not sure where the next bit of revenue comes from. Never mind that the things I was worried about one, two, three, or four months ago have all gone better than I feared - anxiety doesn't care about that. It'll just glom on to whatever excuse it can get - see e.g. https://news.ycombinator.com/item?id=40191179
Keep at it, let the time heal you, I’m rooting for you. I was, and still am, having mental health issues for many reasons. Some are permanent, some are not. I’ve recently begun to be able to engage with hobbies again after at least months, if not years. You’ll get there too, I am sure. I won’t give advice, just a friendly note :)
Me too. I was in a very dark place for a long time until a couple months ago. My views on life and death haven't changed but I'm not actively rooting for my own death anymore, at least.
Embracing the solitude and the mindset that no one is coming to save me has sparked a bunch of motivation (out of spite) to improve myself. Started to workout regularly which has been a huge help for my mind and body. Starting to peel back layers of things that happened during childhood now.
Good for you! It really is the very most important thing, to find ways to not be at war with yourself when the world's got more than enough things for you to be at war with to begin with.
If you ever need someone to talk to (who is also firmly on the spectrum + has dealt with severe mental illness in the past + is mostly on the other side of it), hit me up anytime. Email is in my bio (it's my work email but it'll do).
Sometimes things click and sometimes they don't. Like when someone tells you not to worry about the things that you can't control. What worked for someone close to me was imagining and believing that everything will work out in the end. Over time they have seen the positive in life and opened more opportunities than when they were cynical.
Yeah, I need that too. I was able to get a therapy appointment in October. My last boss at Rent-Seeking Publisher screwed me over past the degree of "people used to duel over things like this". Also other shit.
I'm working on a web scraper to assist myself and a friend in booking courts for our sports activities. It helps book the courts when they are released at 12am. This means we can get back to sleeping at reasonable times. We have a long term booking, but don't always require two courts.
On other things, I've been playing with rustlang. Looking to build something with it.
New computer game for blind and visually impaired people. I have computer prototype, but working on hardware to make it happen. Unfortunately, hardware isn't easy to tackle, I need to create custom keyboard-like buttons.
If you think about games, there is handful of them for blind and visually impaired players. But things got more interesting when keyboard is your screen.
I am working on an agent cognitive architecture based around LLMs. Been working on it for a while. The main difference from all the dozens out of agentic systems out there is that has been designed so that it can introspect fully and it has a machine ethics layer based on the work a modern Stoic philosopher. I am trying to build a system that I would like to exist.
Rediscovering an old love for writing that I thought had left me after highschool, now applied in the field I work in! Recommend starting your own blog if you're on the fence + being consistent about posting, it's extremely worth it
I'm building Kweet, an AI-powered analytics for nonprofit organizations. It allows nonprofits to get new donors by analyzing, predicting and improving their social media.
Waitlist open: https://getkweet.com
Preparing my PyCon IL workshop, a competitive platform where you write code to solve simple games (think snake, pong etc.). Here are some screenshots: https://imgur.com/a/VDKfmjO
After it's done I hope to add a demo to my blog.
Cheers for all the interesting work going on here!
We are working on Digger.dev - The Vercel for Terraform
Digger automates deployment of IaC like Terraform or OpenTofu. You push Terraform code to GitHub, and it gets planned and applied - like Vercel, but for infrastructure code. You can manage environments, secrets, policies and drift in the web UI.
I'm working on a screen reader/narration service for retro games, called groan. https://github.com/JesseTG/groan It's still in development, but I intend to post more about it once I fix some related bugs in RetroArch.
I'm working on updating the https://yavaca.com site including the backend written in golang. It is for users to publish out-of-office (OOO) plans so they can unplug while on vacation. Great for maintaining a healthy work-life balance. Having fun with the golang part.
I'm building a service that helps event DJs engage with partygoers and take song requests over text message. It is SMS based as guests won't want to download an app just to send a request in. https://requestnow.io/
A NES emulator in Typescript targeting the browser. I've been using the Messenger emulator with cycle-accurate emulation to hone it. This stems from my interest in low-level machine code and understanding how computers work. I want to eventually write my own virtual machine and create a functional programming language on top of it.
Free is everything, but runs on the existing system tables; so it's slow, and so a six day history only.
Paid has a system table archiving mechanism, so you have before-and-after when something goes wrong, and where the tables are in Redshift proper, not the leader node, it's fast.
I'm learning the course about SEO. I have never heard of this term until now.
Although some people say the SEO is dead and does not work anymore, I am still spending effort on it and building BMW VIN Decoder https://bmwvindecode.com/ to learn by doing.
Still working on Octarine (https://octarine.app). Recently released Linux images, so more people could try it.
Currently re-writing a bunch of the bundler stuff and how UI takes data in for performance improvements, before shifting gears to building the Windows version!
My friends and I have been working on building our website — https://bigbeans.ai
It's leetcode for ML/AI space.
We believe that a lot of software engineers are interested in learning about GenAI and like us learn by solving problems. :)
Hey, good work. One small feedback, after logging in (from mobile) it took me to the profile section rather than listing problems. I was expecting it to redirect to the problems section. From there, it was tough to find the section. That displays problems.
A LLM based bot for information literacy. The idea is that it can analyze text and images to help folks understand if they should believe stuff they find on the internet. And what they need to know to understand the subject.
I'm making a cyberpunk-like ttrpg system, that should be roughly backwards compatible with the old gear catalogs. But it adds better progressions, some fairly innovative (I think) die rolling systems that make better use of party skills, and a number of ways that roleplay changes the rules, and visa versa.
Building transcription app that lets me dictate notes to notion.
Doint a lot of exploratory work for my day job and trying to establish practice of good documentation so i hope being able to ramble as i work will make this easier.
At this point it's simple macos flutter app that toggles recording on global keyboard shortcut.
I’m back to working on an ML app pair (train/consume) for Mac/iOS, for finding a certain type of resource.
I tried it a while back, and wasn’t too thrilled with the way Apple has structured their ML stuff (they basically only afford very specific applications, which didn’t match my workflow).
Hey guys, I’m currently working on a design handoff solution similar to Relay by Material.io, but with support for iOS, Android, Flutter, and React Native.
The goal is to streamline the design-to-code process.
Would love to hear thoughts from anyone who’s tackled similar challenges or has insights on optimizing cross-platform handoffs!
I’ve built a tool that allows anyone to add an AI chatbot to their website. There are lots of features I would like to add but I’m making the jump into sales and marketing work by starting a youtube channel to promote the product.
Writing an NIH SBIR grant to fund a software idea a friend and I had. The software we’ve developed seems to directly match what the solicitation is looking for, but it’s our first time applying for a grant of this type. My friend is a researcher and is familiar with NIH grants, so that definitely helps.
I quit my job in the spring to work on UI design software where designers visually build real component and design system implementations, and engineers can collaborate by editing the code.
The goal is to eliminate the massive chasm that exists in design-to-engineering hand-off by having a single source of truth.
If you're doing this as a business, you should consider how you think this is different from e.g. Figma (for designers/PMs) or Framer (for design-implemented-as-actual-usable-component-on-a-low-code-site). What you're describing sounds very similar to Figma's developer mode, at least framed the way you frame it here. Is the idea that, say, the design files are loaded as CSS or React components or whatever directly into your site, rather than being copied or imported?
(If this is just a personal project, disregard! But "quit my job" makes me think you're thinking of this as a business.)
i've been working on a webhook to instant messaging service called https://hooks.im after reading about a similar project here on hacker news (i think it was https://news.ycombinator.com/item?id=41144227). you can create webhooks, get a receiver-url that you can send your webhook invocations to and receive webhooks via im (currently only telegram). i've also added a jq-like template engine to reformat the request bodies (if wanted) to extract specific information. (only works for json bodies, obviously)
this is all in super early non-peer-tested alpha (if anything), so feel free to try it out and give feedback :)
We're continuing to develop and refine our core offerings at Elefunc:
RTCode.io is our real-time web development playground for HTML, CSS, and JS. It provides instant feedback - as you code, you see the results immediately. A key feature is our support for in-editor Workers, allowing developers to write and test backend code directly in the playground.
These Workers integrate seamlessly with RTEdge.net, our global, multi-cloud edge network. RTEdge.net offers distributed hosting with auto-scaling capabilities, giving developers a powerful platform to deploy and scale their applications worldwide.
We're also running rt.ht, which we tout as the world's shortest SaaS eTLD.
Our focus remains on providing developers with robust tools for real-time web development and edge computing. We're always looking to improve the integration between RTCode.io and RTEdge.net to streamline the development and deployment process.
If you're interested in real-time web technologies or edge computing, we'd love to hear your thoughts or answer any questions about our platform.
I’m interested in High-Frequency Trading (HFT) using algorithms, starting with basic trading algorithms like RSI Divergence and Arbitrage. I’m utilizing the Binance API to retrieve ticker data. If anyone has experience with this or is interested, I’d love to connect and discuss.
I'm working on a social website where you can share the soft drinks and sodas you try, similar to untappd and maybe swarm, it's not yet ready to share, but I'd love to know if there's any people here that might want to try it (my email is on my profile)
Stenography/text expansion app that emulates a Bluetooth keyboard. Combines regular character entry, stenography, and fast autocomplete using an N-gram language model (lower latency than an LLM, more transparent/debuggable, no need to cap vocabulary size for embedding).
I’ve been slowly adding puzzles to a little game I made. Shout-out to a few HN commentators who gave early and helpful feedback. https://shovemedia.com/letters
First-time founder working on a B2B comms web app. Have had trouble locking down technical co-founder but recently made great inroads with a childhood friend of all people. If an MVP comes out of it, I'll post it here!
Im working on a system/formula that rates and organises human expressions/ideas/insights finds and organises relevant data, solicits more expert reviews, forges presentation for limited attention spans, develops a plan of action and a cost/benefit analysis.
I'm working on a side hustle to help technical consultants build newsletters. Makes me sad when I see a great technical answer on a slack that I know will help far fewer people than it could.
So this weekend, I'm building some processes and marketing materials and doing some outreach.
A stock price analytics webapp. That gets data from various sources. And is able to bulk analyse historical data. Made for a friend. But now used by more peeps.
Working on generating zsh tab autocompletion for my CLI framework. I'm finding it difficult because zsh makes completion surprisingly complicated and the zsh docs have a lot more descriptions than examples. Other CLI frameworks all seem to approach this differently.
Building a toolkit to save hours of time when working with LLMs locally. For example, adding a TTS or Web RAG to the Open WebUI setup both are a single command.
I've been working a lot of some games in Python and Panda3D. I want to make somethingusing the joy cons as motion controls. Also been looking into stackless Python for a text based multiplayer game I'm working on.
I need to update my website at some point its been a while.
A huge sports LED ticker screen. It’s been an interesting experience. Hardware is hard. But people love it when they see it and seem to be really excited for it.
A speech-to-speech translation app, with automatic language identification. One touch interface to start/stop. No configurations required for everyday use cases.
That's the plan... :-) It can auto-detect 45 languages now, including Chinese, Japanese, Arabic and most European languages. (like in sci-fi movies, it allows two users to share a pair of airpods to talk to each other in different languages.)
I'm working on a Firefox extension that analyses the user's browsing data. It tracks visit frequency and time spent for each website. I'd like the extension to generate a distraction score as well, depending on these two data values.
I made https://rascalhq.com, just to experiment with embeddings and because I wanted to see what kind of “advice” Shakespeare and Warren Buffet would give me for various queries.
Working on a ton of new music - https://4D4M.com and 4D4M on Spotify.
Currently exploring production of new genres and some hybrids as well as some tech to improve live shows.
For the last decade or longer, I have been using pinboard.io, and I wanted to add a few more features. I just launched the landing page, and that's a start!
I'm putting qnx cross-toolchains into Debian packages because $DAYJOB builds qnx images in ubuntu containers. I've learned a lot about cross-toolchains and my appreciation for blackberry's technical prowess has never been lower.
An open source alternative to Strava (a fitness tracking app that allows users to share their running, cycling, and other workout activities using GPS data).
I’m constructing my house alongside a small crew, who are guiding me through the process of welding, erecting brick walls, and laying porcelain tiles. While I’m definitely slowing them down, it’s been an incredibly rewarding experience.
I'm working on updating the yavaca.com site including the backend written in golang. It allows users to publish an Out-Of-Office (OOO) plan so they can unplug while on vacation. Great for maintaining a healthy work-life balance.
Trying to make my own thing that transcribes what I hear throughout my day. A lot of it is just finding an efficient stack (eg choosing silero instead of pyannote, azure vms instead of apis, whisper.cpp instead of faster whisper)
(ha ha) no really, an app that aims to help out older, lonely, but good people. It's a desperately underserved market. These people are often overlooked and we're trying to empathize with them via software.
I'm working on a platform that allows students to view and work on relevant past exams from anywhere, gamify and track their scores, and maybe add some LLM integration to discuss answers and ask follow-ups.
Been working on a digital time capsule app for my friends birthdays and other big life events. I think there’s always opportunity to share more moments with people, so I’m hoping I can use this and get my friends to use it too
I'm working on building an API for getting AIS data for ship tracking. I built an app a few years ago for tracking local ferries and getting the data was a nightmare, so I'm trying to build a simpler alternative.
Leetcode for a new job, been at my current place for 5 years. But it's actually the perfect thing to do to make your brain come up with new projects that your brain would rather do, so I write those down for later
I'm actually creating a new web framework/library. I know. I know. Yet another one. I got sufficiently excited about the technology recently that I've dived head first into making one. Stay tuned.
Re-launching a job board and social for a niche highly regulated market. Built mvp years ago, now finally refining it and exploring ways to monetize. cannabisverified.org
Drupal interface for interacting and managing ArcGIS Online services, layers, and tables for people who understand GIS but aren't GIS experts to build reports that query across potentially unrelated services.
Despite not liking JavaScript as a language nor the Node runtime, learning about hybrid static and SSR apps with Astro.js - I have to say, I’m impressed so far. Seems like a sweet spot for certain types of sites.
On JumpHigher [1], an app that helps you improve your vertical leaps, just using your phone camera. It's functional already, but very early, the algorithm took quite a while to nail ( still far from perfect ).
I'm a CEO of a small consulting company, and I love working with startups ( or hate working with huge companies, matter of perspective ), and I always thought that to be better at that we have to be able to launch our own stuff.
Also, I never launched anything solo, working alone is hard for me so this was a challenge. I want to continue working on it on the side for a while.
The main audience for now would be basketball players that want to dunk, or anyone that wants a good leg workout and track the progress.
A .NET MAUI travel app that uses GTFS data to present transport info to users and also includes interactive 3D graphics.
Not on Play Store yet because the app is too big, still figuring out how to solve that.
A golf swing measurement app using 3 sensors and an esp board. Just started practicing golf and made sense to understand how the swing works. Claude AI is helping me with some of the coding, fun stuff.
I am creating a RAG trained with HN post "Who is hiring".
It's nothing special, but it's just a simple project to get confident with LLMs and prompt engineering
been my main side project since like 2008. working on goodreads import right now. Always working on improving the algorithm. would love to collab with a data scientist on ways to improve my algorithm. https://github.com/ssherman/weighted_list_rank
I'm working on a single-file executable version of Fivetran you can run wherever you like, for free, without needing an account. It may or may not be OSS, we'll see.
We have over a 100 places now, and are getting ready for closed beta in September. You can sign up by emailing hello@doggable.app or DM-ing us on Instagram @doggable.app.
I am developing a Pythonic alternative to ArgoCD and FluxCD that uses no CRDs or in-cluster controllers for continuous delivery of applications to Kubernetes clusters
I’m working on building a payroll outsourcing company targeting the construction industry. 2,900 employees paid last week! Deep ERP integration is our competitive edge.
Compared to others, its goal is to make best practices easy, be declarative, robust and fully featured.
I’m a self hosting and data sovereignty advocate and this project is my contribution to this space.
Best practices easy is made by adding a layer on top of the stock nixpkgs modules that make for example Vaultwarden easy to secure (/admin behind SSO with Authelia) and easy to backup. One doesn’t need to understand how to setup Authelia nor what folders to backup. That is taken care for you. https://shb.skarabox.com/services-vaultwarden.html
Robust thanks NixOS being a declarative OS, for example by adding a grub menu for every new deployment, making rollbacks easy. But also thanks to extensive tests that for example validates that the SSO integrations do not break on upgrades.
Fully featured because for every service it provides, it makes setting up the reverse proxy, backup, LDAP, SSO, etc. easy and importantly in a standardized way thanks to contracts I’m adding https://shb.skarabox.com/contracts.html
All of this is not meant to stay in this project though. I’m slowly working towards upstreaming everything into nixpkgs. I’m starting with upstreaming a feature that allows out of band secrets to be interpolated into config files easily https://github.com/NixOS/nixpkgs/pull/328472 I make extensive use of this is Self Host Blocks already.
Finally, I’m using this project for my home server and one in my parent’s place. So I’m using it in “prod” already.
An app for sharing photos with a large group of people while still preserving the privacy of the people in the picture who don’t know each other.
Our kid’s nursery school posts a lot of photos to Facebook, and we’ve asked them not to post photos of him. They’re (mostly) respecting our wishes, but we still have a bit of FOMO over the pictures of kids having fun, and wish we could see pictures of our son having fun too.
The goal is that the nursery would be able to upload photos and parents can see only their own child’s face, other children are pixelated.
Still early stages but got a working prototype, and I’m enjoying the building process.
Working on an advertising and marketing strategy platform with my brother comprising a series of apps that accelerate workflows and time to insight using AI.
I've overlooked the importance of having a good virtual presence, I believe it's better to be noticed and be an average developer than being very great but working alone. The former is more impactful and not that I intend to be average but I'm starting and would love to have a long impactful journey. so focusing on having noticeable social handles.
I didn't like web dev very much but now I'm enjoying Django a little so I intend to swallow Django and pull off some products maybe as I don't intend to have a regular job for the rest of my life and build what interests me.
So yeah, I've been pushing some Django projects and getting better at it.
If you wanna follow along or learn Django, here lies my journey
I'm working on an end-to-end encrypted alternative to Google Forms (because Google is creepy).
It has all the same features plus many more: advanced conditional branching, webhooks, thorough design customisation, better question types (e.g. address autocomplete), etc.
I've bootstrapped it into a (very) small startup. The idea is to maintain an ethics- and privacy-driven form builder that people don't have to trust, since nobody can see your responses. It also makes GDPR way easier for you!
Thank you so much for the kind words about TextSniper! I'm really glad to hear you find it useful and use it frequently. It's always rewarding to know the app is helping people in their day-to-day work.
I’m working on a pretty cool private equity platform at Pactio.
It’s probably the most complicated product I've ever worked on. The team is incredible & I’m lucky to be surrounded by incredibly talented people. Our head of engineering is a fantastic person too, so it makes working here so easy!
In my spare time I’m working on an AI assistant. I thought I would try sell it first, but given the ramp uptake I’ll probably open source it.
It boils down to simplicity. While error reporting is extremely valuable for most projects, it is absent from many. The most common reason is that this requires entering SaaS territory. Most solutions are provided by third-parties that require you to subscribe and pay for the privilege of tracking your errors. Cost is the main noticeable downside. Data protection (GDPR, HIPAA, etc) is another.
Super polished looking product -- any success in getting companies to use it? Work at a dinosaur that would never think of abandoning the microsoft ecosystem; wondering what the target is for something like Loops.
(I did try taking a look through the website and saw integrations posted more prominently than users?)
Hey. I’m building https://www.unwrangle.com to make it easier to scrape data from ecommerce and other sites that are hard to scrape.
It’s like Apify but my goal is to make it easier to use
Currently, I have APIs and scrapers that work with platforms like Google Maps, Yelp and Amazon. The APIs are useful to get data immediately and scrapers to extract information from many URLs.
The plan is to add more general purpose APIs like an HTML API, Markdown API and eventually features to build your own api and scrapers with AI.
There’s a lot of tools in the space nowadays but imo they are all flaky. My intention with unwrangle is to offer a way to scrape any site that just works without the need for any config or complicated pricing. The project is at a little over $1000 MRR. Marketing it has been and continues to be a big challenge. I’m bootstrapping solo and hoping to reach 5-10k MRR in the following months. Plan for that is to consistently improve offering and conduct marketing experiments.
What I find interesting about it: I’m offering easy ways to scrape sites on which antibot is really hard to bypass like Twitter, paywalled sites, LinkedIn, etc. The ability to build crawlers without writing any code is kinda cool. User who would normally not have used web data are scheduling scraping jobs and using the data for analysis. For the HTML API I’m thinking of doing an interesting spin on what others like ScrapingBee are doing and abstracting the needless config like premium proxy etc. and just effectively offering a higher # of requests. Also for the build your own scraper, offering users a way to create a parser with a prompt and use it in a single browser session to collect data from many pages saves hassle compared to a synchronous API approach
Im working on yet another "what's my IP address?" site: https://ipparrot.net for the fun of it. It's written in Elixir and currently working on adding a structured whois tool and plan to add more networking tools. You can also use curl like:
godot csharp (C#) game engine tech / middleware. If you are in the seattle area or want to collab on hobby stuff with questionable financial upside, ping me.
Made it to the front page of HN a few months ago with a less polished version. Still trying to build consistent traffic and engagement. Any feedback is much appreciated!
I have apps that pay my rent such as
YOU-TLDR - Transcribe and Summarize YouTube videos
you-tldr.com
Shorts Generator - Text to Automated Shorts In Minutes
shortsgenerator.com
Snoop Hawk - Automated Web adn Reddit Marketing
snoophawk.com
You can find all my projects here
hackyexperiments.com
Currently I am exploring the idea of chatting with more than one AI. My assumption is it makes the interaction more life like and less lonely and I think in the future everyone will hav a group of AI friends. I made a quick loom about it here: https://x.com/deepwhitman/status/1826831221554643324
I’m writing a series on how to understand golang if you’re an experienced TypeScript developer.
I’ve worked with several teams now who wanted to adopt Go for the backend, but the team’s only extensive type system experience is in TypeScript. They tended to hit very similar pitfalls and want to coerce the type system to behave like what they’re familiar with, and inevitably they wound up with inflexible and error-prone architecture.
They also failed to accomplish any kind of holistic, 30,000ft view architecture because implementations were too often sent off the rails by not understanding the language deeply enough. That was often in part due to type system confusion inherent to how Go implements certain data structures (particularly slices), which leads to the sprawling repetitive Go code people dislike so much.
I did all of this too, so I feel like my insight and experience would be valuable to share. I actually wrote 90% of it last year and felt like a goof, like, ahhh what am I doing obviously everyone else figured out Go faster than I did and this writing is just telling people how incompetent I am. Haha. But then I worked on another project and the situation was exactly the same. I actually used my draft content to help the team understand how I see the problem, and how to get past the mental traps that they were in.
Fundamentally it comes down to “learn the type system/language properly”, but it’s not quite so blatant or unhelpful. One of the problems with TypeScript methodologies is that they’re so ridiculously flexible, but go is not at all in the same ways, yet they have just enough superficially in common to trick you into thinking otherwise.
So, my series essentially elucidates the similarities, differences, and what’s totally absent in Go from the bottom to the top. It’s interesting because they really are remarkably similar in some ways, which is precisely why it can trip newcomers up. Knowing why that’s a trap makes you a far, far better Go developer. Along the way it also reveals details about TypeScript that some people might not be fully aware of (at least several in the team I shared the drafts with said so), which could be helpful.
I mostly need to muster the courage to hit publish now. My confidence has taken a beating in the last year or so!
I'm a Flutter developer starting to be a bit disappointed in the tech (and the community) and sceptical if it can really produce great apps, so I'm thinking about switching to something new in the coming years.
I decided to build a multiplatform app with SwiftUI to test out the waters. So far, learning SwiftUI and building an app is somehow both easier and harder than expected, but I'm glad to see something new.
I'm planning to write a series of apps in the next 6 months, so that I build confidence shipping apps with SwiftUI and maybe find a smaller contract with it.
It's certainly not terrible, and I'll probably stick with it a year until I learn something else as well as I know Flutter...
But to answer...
Main reason, I'd like to ship software that isn't only good, but great. To create great apps, it needs to integrate into the OS/browser, and my experience with Flutter has been not great in this regard. The apps always look and feel a bit foreign. I'd rather write 3 separate apps (web, ios, android) that feel great, than ship something faster (if even...) that is so bad, sometimes I feel it's a spit in the face of our customers.
The mobile alternatives became much better: when they started, they were (as I remember) significantly better than the alternatives. Since then, RN became much better, Swift and Kotlin became mainstream, SwiftUI, Jetpack Computer came out, and in a couple of years KMP will also offer a pretty compelling alternative.
Flutter is spread too thin: I understand why they did it, it's great for marketing, and I know people who use it on web, desktop and embedded and are happy... To me, however, it feels like they added platforms before making iOS and Android work and feel great. I hate that I had to debug and pin dependencies that broke due to web incompatibility on an a mobile only app. IMO, it's so shjt on web it's laughable.
I am a mobile developer using Xamarin/.NET MAUI for some years now. It is not as popular as Flutter but im pretty happy with it. In my company all of our apps are Xamarin/MAUI too and these are used by hundreds of people.
MAUI is not perfect, theres still a long way to go, but it is getting better.
We still using Xamarin because of that.
I'm trying to push the boundaries of SQL by building Anyquery[1]. It's a SQL query engine that allows you to run queries on anything (GitHub, Todoist, Parquet, Google Sheets, logs, emails, etc.)
It's mental gymnastics to transform different data sources (e.g. a spreadsheet) into a SQL database with write support, but I do enjoy the journey and learn a lot from it.
Open source app to download emails, provide fast search, offline access, automatic labels, categories, summaries, surface up what needs our attention, etc. (some features need AI).
Starting with emails, calendars and attachments. Integrations for Slack, LinkedIn, Stripe, databases, even web crawl are coming...
are you going to cover the whole process end to end, from my cash to my friends cash assuming neither of us have a crypto wallet yet?
what about transfer fees?
all in all this is way to complex for the average person and inaccessible for many.
in some countries mobile money accounts come with every phone number, and you can transfer micropayments from one number to another even across networks and even internationally. and you don't even need a smartphone. any feature phone that can send USSD codes will work. that is a working micropayment system accessible to anyone who can afford at least a feature phone.
It's way too complex. But it's possible. I'm in the US, my site will be targeting US users. if you want to do something for EU users with USSD, I'd love to collab, email in profile.
ok, it is reasonable to focus on the situation in the US.
my point about USSD codes was that a globally applicable solution for micro payments does exist, and using a complex system with many steps really just begs the question of what prevents these simpler systems from being adopted.
the answer is probably politics and lethargy. the inability or unwillingness for big companies to innovate.
to be clear, i am not criticizing your work, but the state of micropayments in the US, and i am mostly interested in discussing what can be done about that.
showing people how it can work now may or may not be part of that. personally i am more inclined to look at this process and run away screaming than make any attempts to adapt it.
yeah, fair. I'm expecting most reader's eyes to glaze over so I'm expecting zero money out of this, given all the hurdles I hit to accomplish this task that should have been easy. but the first step is to document the current state of the world and get that out there.
My wife and I are building Spruce, a Slack app that autocorrects typos in your messages: https://spruce.so/
The motivation behind this was twofold. Firstly, we're both the type of people who often edit our Slack messages (or any other messages like WhatsApp) to correct mistakes. Secondly, my wife is a PMM at a tech company, but she can't code. I've been telling her that anyone can create a decent product with the help of LLMs and tools like Replit, but she usually just rolls her eyes. So, on a flight from London to SF, I suggested we build Spruce. The idea was that she would take the lead and rely on LLMs (ChatGPT + Ollama for when the wonderful Delta internet wasn't working), and she could also treat me as a [more intelligent] LLM. After ~12 hours and a few small arguments, Spruce was born. :)
Some random learnings, top of mind:
1. A non-technical person, even with current LLMs (GPT 4.5), can't go from idea to shipped product without at least some help from a technical person.
2. I now make even more typos because my habit is to type as fast as possible, often without even pressing the space bar. Spruce autocorrects everything in Slack, but when I'm using other products, I feel incomplete.
3. Although I'm practically a native English speaker, it's still my second language. I've added a feature to Spruce, just for myself for now, that DMs me if I make really obvious grammar/English mistakes in my messages. I decided to do this after one of my colleagues, who is very English, said "for prosperity" instead of "for posterity" in several messages, and another said "for all intensive purposes" instead of "for all intents and purposes." It's been a very cool development, at least for me! It's fun because a) it has a nice tone, and b) it's not verbose and only nudges you on major mistakes, and only up to twice a day.
Anyway, feel free to give it a try: https://spruce.so/
P.S. This Slack app necessarily needs the permissions it has because it needs to be able to read and edit your messages everywhere (e.g., public channels, private channels, DMs, etc.). Don't install it if that freaks you out. However, it's easy to remove Slack apps, so there's no harm in giving it a try.
Working, well, ideating, on an embedding solution (read blatant scraping) for various kind of forums and then hybrid search + LLMing over it. Need to devise a UI as well in lieu of the new gpt chat modalities.
I'm working in the final features for an office productivity suite with deep AI integration. I think the current slapping of a chat window on software is pretty tame, if not lame. So, in addition to that chat window that replies with conversational chat, I've got over a dozen integrated into software tools LLMs that when you ask them things, their output is programmatic and either modifies the data of what you're working on, or or performs some type of modification to the software tool in use. The tools include a complete word processor, where the AI knows the word processor's API and can directly manipulate the document in the word processor, likewise for a spreadsheet, and there is a "prompt engineer" interface where one can clone, modify, and create new LLM Agents that have deep access to the APIs of the tools, the data in the tools, and the wider application framework.
This has been in development for 3 years, and I've got an immigration law firm using it, with about 1/3rd of the LLM agents being immigration law specialist agents of some type.
I'm just wrapping up transforming the system from being immigration law specific to being generic, capable of operating in any industry. I previously made a "do it yourself: build your own home solar energy system" as a proof of concept that this framework could be modified in such a way, which I put online for a few months and then took down, having proved what I wanted.
I've got multiple creative writing 'bots: legal, technical documentation, creative writing, and code authoring. I've got multiple spreadsheet 'bots: create any standard spreadsheet form on demand, reverse engineer and explain complex spreadsheets, and co-author spreadsheets interactively with the human user, guiding them through the understanding of the spreadsheet being built. I've got foreign language translation agents that allow people without a common language to speak to one another through the voice transcription interfaces.
And the users are never copying and pasting LLM outputs from one place to another, that integration is built in to the business logic of the software: ask a chatbot to write a document, the output does directly to the word processor and the document is created, likewise for spreadsheets, likewise for talking to the "projectBot" and asking "what's the state of this project?" and a detailed report is generated.
I've also been making the app itself multi-lingual, and multi-skinned so it can be refaced for different cultures, demographics, and industries. I've been calling it "AI CMS" but that is meaningless to far too many. I'm considering calling it "Midom Office AI" because that sounds like "my dumb AI" and I'm generally sarcastic, considering an anti-gushing sarcastic marketing angle on the software. Rather than everyone's else's over praising, I'll have just some confident smuck referencing how he's got an entire team of AI experts helping him, enabling him to be calm and cool in the face of all the deadline pressures, he sips lemonade while his AI team works for him, and not him for it. We'll see what my "marketingBot" says...
I've been building local tracking for job search listings[1] in my spare time, along with crawlers[2] as data sources. Mostly in bursts though, bc I've been juggling fleshing out the project with actually using the data. (On a related note, the current working state of the backend hasn't been pushed in a while, and still reconsidering the name.) It's mostly been an API up to this point, and most recently I've been chipping away at a frontend to accompany that, which is beginning to be useful on a personal level. In particular, I just got file upload working. But long-term, I see it as a hybrid between Tinder and a CRM (and I already have "swipe" animations working if not the actual touch handling), where you can bring your own data on jobs and companies -- however you get that, because I'm sure you can find better scraping processes -- and have a way of organizing them.
It's a lot of experiments. On the frontend, I'm dispensing with any design systems or styling libraries, and have opted for vanilla CSS and hand-rolled animations. I'm using React for now, just to get it scaffolded faster with tools I know already, but I've been eyeing a few others, for reasons that range from a standing curiosity about Web Components to new ones about signals. The backend stack is all Deno-based: it includes Astral for any scraping needs you might have, uses Oak for routing, and leverages kvdex for storage. (I also contributed `model.getOne` and `model.updateOne` to the latter as part of initially getting duplicate detection working.) My goal is to keep the whole thing light on external dependencies, outside of a few pragmatic options, and generally work more directly with platform or runtime (or in Deno's case stdlib) functionality where feasible.
I otherwise rebuilt my website over the last few months, using Lume for static site generation with similar aims about shipping something generally lighter (the content there is all Markdown, and so far, everything but the syntax highlighting and the ToC generation are built by hand). Additionally, I've been using it as a home for writing[3][4][5] and not just code, so that I have some other things I can show off. (The last one made the front page a couple weeks ago!)
I am working on a strategy game. I call it The Lost Age Of Abundance. A multi-player, real-time war-and-economics game. It is the kind of game that most developers would typically take a turn-based approach to, but I think turn-based games are mostly boring, so in this case it is real-time. My plan is that at least 100 players can play at once, but perhaps I might be able to get that up to 1,000.
There are some unusual things about the game. To win, you have to conquer all of the land. However, no player is ever knocked out of the game. If you lose all of your land, you can simply become a guerrilla leader and engage in guerrilla warfare, in the hopes of turning your situation around. Being a guerrilla leader is difficult, but if all of the guerrilla leaders form an alliance, they might have the strength to reduce the power of whoever is winning at that moment. Every player starts off as a Noble Lord but you can become a Priest or a Communist or an Anarchist. (The Anarchists have a unique way of winning, which does not involve conquest.)
I have been working, for months, on the economics, which are very complex and I think very interesting. All prices start off randomized (within certain bands) so there are no "good units" or "bad units" only those that are under priced or over priced. There is an economic cycle that runs once every 60 seconds. It takes all purchases and turns the money spent into inflation. Specifically, the percentage of total world GDP that was spent on a particular product is the inflation rate, so if all of the players collectively spent 2% of their income buying swords, so 2% of total world GDP was spent on swords, then the price of swords goes up 2% for that minute. In this way, with players buying what is under-priced and thus driving up prices, the prices are pushed to their rational levels. Also, the inflation rate tells you something important about what other players are doing. If you notice the price of horses is going up at an alarming rate, it means one (or more) of the players is spending heavily to build a large cavalry force. The economics are entirely deterministic, based on the activity of the players, there are no random events. There are no randomly generated earthquakes or floods or NPC invasions. Every change arises from the actions of the players.
I was originally going to give the game the historical setting of the Chinese Warring States period, but then I decided to go with more of a fantasy theme. But there is nothing overtly fantasy in the game, it's simply that there are many myths, which the peasants might genuinely believe, that explain how the Empire Of Abundance was transformed into the Ruined Lands. The peasants are looking for some leader who can restore the Lost Age Of Abundance. The peasants (NPCs) will switch their loyalty to whoever they think can win. Occasionally some peasant will become a True Believer, in which case their loyalty is fixed for the rest of the game. That means even if their chosen leader is defeated and becomes a guerrilla leader, the peasant will remain loyal to them, and secretly send them small amounts of money, to help them recover.
If your peasants think you are doing a terrible job with the economy, they will emigrate to the lands of one of your enemies. You can seal the border, to stop the emigration, but the peasants treat this as tantamount to putting them in prison, so their morale will plunge. If their morale gets low enough, they will switch their loyalty away from you and to any nearby guerrilla leader.
There are 3 games in this game:
1. the military game
2. the economics game
3. the spy game
The military game and the economics game are entirely deterministic. At no point does the software "roll the dice". Combat is as deterministic as in the game of chess.
The spy game does use probability. If you torture a peasant then there is an 80% chance they will tell you the true. If you are wondering if one of your peasants has become a True Believer for one of your opponents, you can torture the peasant to get them to confess. There is an 80% chance that whatever they tell you is the truth.
I have almost zero graphics for this game, so I think it will only appeal to a limited set of people who are interested in a game with a very dynamic economics system. The point of view is basically that of a ruler sitting on their throne in their capitol -- they only hear the reports brought to them by messengers. They cannot see the economy, they can only read the reports and graphs given to them by their court scribes. Likewise, with battles. In that sense, the whole game is a bit abstract, but I hope it will appeal to some niche that wants something complex.
Your game reminds me of some ideas I've kicked around, with players being nobles of some type, and having the same point of view (receiving updates and sending orders).
In my game these player nobles would functionally act as kings/queens, governors, generals, admirals, or travelling mooches, depending on their goals and how well they are doing. The map would be broken into cites and towns arranged into territories, nothing super granular. I wanted to feature politics, where players would normally need to interact with each other, with shifting alliances, etc. A general might decide not to obey their king, perhaps planning a coup with some governors for example. Kind of like the board game Diplomacy. A king would need to coordinate tax incomes from governors, dispatch generals and admirals, expand alliances, and be wary of threats. Governors could influence trade and improve their territories, cities and towns. Generals could perhaps take on mercenary jobs, or raid. An ambitious admiral might want to take a city and move up in the world, or be happy enriching themselves in trade or pirating, etc. There would be gold and various other resources, population morale, seasons, spies, etc.
I was going to have this game be turned based, with each turn being a month of game time, and players only being able to take one turn per day of real time. The world would be one month in diameter - so that a fleet or army or inter-player message could be dispatched to anywhere in the game in a single turn. I figured that this would avoid the possible problem of players gaining an advantage by being able to coordinate outside the game, and to keep the programming as simple as possible. A minor character might not have a lot to do on any given turn. A king or queen could be a very busy person. Players might be on autopilot for some number of days if they don't log in. News in the game would travel also, so any player could hear about most of what was going on in the world. I shared your sentiment on graphics, figuring I would leave room to add those later.
I like your ideas for your in-game economy, and what you described about the effects of morale, and true believers, which I'm sure dovetails with there being priests. And it's cool you have a place for anarchists. They get a bad rap.
"The world would be one month in diameter - so that a fleet or army or inter-player message could be dispatched to anywhere in the game in a single turn."
I had a similar idea. Since my game is real-time, rather than turn based, I decided to use time for to indicate travel. There is basically no map. In that sense, it is similar to some sci-fi star games where star travel is mostly handled by "teleporting" from one star to another. Though I use time to indicate that attacking another Kingdom cannot be done instantly.
Thank you for the suggestions about Show HN. Good idea. I'll do a Show HN when it is ready.
I'm working on Minders [1], a journaling/notes app. I've been using it for a little over a year for daily journaling and keeping track of interesting links or ideas. It works like a Twitter feed of just your thoughts.
It's been fun to work on as I can throw in whatever neat ideas I want since I'm not trying to fit it into a neat category. For instance, I was experimenting with a multi-column browsing experience after I saw a post here [2] for inspiration. The UI could work like TweetDeck but for your notes. Not 100% sure yet if it works, but I'll be doing more prototyping.
Right now I'm revamping the sync system and other runtime parts as I'm realizing that once you start amassing lots of posts, it gets bogged down a bit. I probably could've architected some of it a bit better at the start, but I tend to try to release stuff as early as I can to force myself to ship and to see what people think.
This is great; how does this compare to Day One? I've been using Day One for multiple years (1000 day streak) and was wondering if I should stick with this. It does what I want but it seems like it's lacking in product development post acquisition compared to other offerings.
Thanks for the interest. It's definitely not at the level of polish as Day One yet. I used to use Day One years ago when it was still a one-time purchase app, but even back then it had nice things like recording your location and local weather conditions. I'd love to add simple niceties like that eventually.
I'm trying to position my app less like a traditional journal and more of a "jot down quick thoughts" place, hence the social media-like interface. But I also want to add more zettelkasten-like features as well, so it's kind of a shoebox app for whatever you want to keep track of for later.
Hi! I'm working on https://UpdateMaker.com. UpdateMaker is a super-simple widget for delivering in-app updates and notifications to users.
As a dev, I always want to tell users about our latest updates, but making front-end updates requires a whole production deploy, and a backend notification service is a lot to set up. UpdateMaker makes it easy to manage updates without an engineering cycle.
UpdateMaker lets you copy-paste our widget into your website, and then you can push update notifications to your users. You can customize the look and feel, schedule updates, track user engagement, and make it conditional on pages the user might visit. We manage all the cookie-handling so the user only sees updates they haven't seen before. It's been super useful for us!
From there, UpdateMaker also automatically exports all your updates as a changelog that you can publish for your users or SEO :)
So lately I've been working on a "v2" that exposes a full superset of GLSL, so you can write arbitrary shaders -- even foregoing SDFs altogether -- in a high-level lisp language. The core "default" raymarcher is still there, but you can choose to ignore it and implement, say, volumetric rendering, while still using the provided SDF combinators if you want.
The new implementation is much more general and flexible, and it now supports things like 2D extrusions, mesh export for 3D printing, user-defined procedural noise functions... anything you can do in Shadertoy, you can now do in Bauble. One upcoming feature that I'm very excited about is custom uniforms and embedding in other webpages -- so you can write a blog post with interactive 3D visualizations, for example.
(Also as a fun coincidence: my first cast bronze Bauble arrived today! https://x.com/ianthehenry/status/1827461714524434883)