Mainly system level and higher, but a bit of all three, I suppose. I was helping reverse engineer a customized SH chip and ended up implementing a small VM and optimized system libraries/utilities afterwards. Most of the time was spent in assembly, with some machine code and C on either side.
It's not really too unusual of a vantage point. SH was clean in some ways, but delay slots are annoying for anything that speculatively executes (which sets a bit of a performance ceiling without a lot of complexity), and more registers are generally better.
