I design and build algorithms and infrastructure for recursive self-improvement, especially around generative models, reinforcement learning, and software-engineering agents. I try to approach research, engineering, business, and prioritization the same way: make things as simple as possible, but not simpler; quotient out accidental detail, preserve the invariant, then build from there.
More context: personal site · writing · resume · LinkedIn
- Top 0.56% in Google Code Jam 2020 across 96,000+ registrants.
- Won ~USD 9k at ETHGlobal Paris for zkAlpha.
- Open-source contributor to vllm, sglang, slime, Harbor, AReaL, prime-rl, Tesla's fixed-containers, etc.
- Perfect 4.0 GPA at Georgia Tech, focused on generative modeling, foundation models, reinforcement learning, and efficient ML.
- A+ in theory-heavy courses including advanced algorithms, advanced linear algebra, quantum information and computation, and modern cryptography.