⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀
01101010 01100001 01100011 01101111 01100010
01110000 01100101 01100001 01101011 01100101
i like deep learning & chips
doing
- gpu architecture at apple. maximising performance-per-watt.
places i did things
- studied cs & ee, researched efficient chip architectures, at imperial college
- interned on gpu architecture team at apple
what really interests me
- deep learning
- computing paradigms & architectures
- co-optimising software & hardware
projects
mSight: a terminal-based performance monitor for apple silicon [1]
tinyflash: a minimal implementation of flash-attention [2]
tinyoptimizer: a minimal implementation of a superoptimizer for tensor programs [3]
research
Asynchronous Arrays: Beyond Systolic Arrays for Sparse DNN Acceleration
writing
how to learn
github
linkedin
X