whoami

I build GPU infrastructure tooling and ML pipelines — benchmarking AMD MI300X accelerators, optimizing LLM inference throughput, and shipping fine-tuning workflows on ROCm. On the side, I write security tools and hack on Web3 protocols.

Current focus: making AMD GPUs a first-class citizen for LLM workloads.

stack

Stats

_{From benchmarks to production — one commit at a time.}