pawan kalyanパワン
i'm pawan. i build database and inference stuff and break it a lot. this is where i write down why.
- good at
- ml infra, databases, inference
- languages
- rust, go, typescript
- elsewhere
- x, linkedin, github, pawann.dev
writing 12 sheets
- 12.kasumi, a few lines of math turned into a cuda kernel19.11.26
- 11.a bias that does nothing broke pythia in bf1611.11.26
- 10.a diff engine in assembly, and a viewer that opens a million lines08.11.26
- 09.running an adreno 630 in numpy01.11.26
- 08.why speculative decoding changed my output26.10.26
- 07.how i made a 27b ternary model 3.8x faster on a laptop cpu by reordering its bits17.10.26
- 06.pointers before cuda: what they are, what changes between machines, and where cuda uses them15.10.26
- 05.the inference tax04.10.26
- 04.gradient descent is one line of code. here is everything that line hides.03.10.26
- 03.your ai has amnesia, and a bigger context window will not fix it10.08.26
- 02.ciot, a cpu inference engine for ternary networks26.05.26
- 01.a membership system that can say "i don't know"24.10.25
pinned to the sheet 32 in the gallery
books, free 2
Below the code
A working engineer's guide to assembly, from bits to GPU kernels: RISC-V, x86-64, SIMD, PTX and SASS.







