writing
reflections on software engineering, deep learning, and everything in between.
optimizing matmul on a ryzen 9950x
0->103% of openblas, featuring the latest avx512 instructions
read article
building a javascript interpreter from scratch
walking through how i built a javascript interpreter from scratch with a lexer, parser, and executor.
read article
building a neural net from scratch
training an mnist classifier in pure numpy, no tensorflow or pytorch, with a 1000x speedup from vectorization.
read article