Ace-2504

C++ and Python are my strongest; comfortable with Node/React/TypeScript too. A systems-leaning generalist still figuring out where I specialize — currently leaning into ML systems and PyTorch..

Currently learning inference optimization (KV cache, speculative decoding, PagedAttention), RAG internals, and CUDA — by building small LMs from scratch and seeing what breaks.

Right now: a from-scratch 125M language model, a tiny dependency-free agent harness, and GPU inference benchmarks (KV cache, speculative decoding, PagedAttention). All on GitHub

Still learning, so I'd genuinely love feedback — say hi if you work on inference, RAG, or systems code and want to tear my benchmarks apart. Open to open-source collabs and internships.