Show HN: I built a 2nd-order PyTorch optimizer for LLMs that runs on 16GB GPUs2 points·by dnosoz·2 months ago·4 comments
How do you design a multi-tenant system without overengineering?2 points·by dnosoz·3 months ago·0 comments
Drarpa – CDC sync engine with exactly-once delivery and saga orchestration (Go)github.com1 points·by dnosoz·3 months ago·0 comments
Next.js template for US credit card affiliate sitestemplate.creditpilotusa.com1 points·by dnosoz·3 months ago·0 comments
dnosoz·2 months ago·discussAuthor here. Happy to answer any deep-dive questions about the CUDA implementation or the Kronecker factorization math.