Well, as long as the operations you want are add/sub/mul/div/array indexing, you don't need to learn LLVM IR—you can just copy the class in this blogpost and write your completely unrelated algorithm in plain Python just as this post does, and it will work.
Basically the author has written a tiny framework for making a reasonable subset of Python a DSL that targets LLVM, which is super cool.
Basically the author has written a tiny framework for making a reasonable subset of Python a DSL that targets LLVM, which is super cool.