About Me

I'm Jin Pan, a second-year MS / PhD student in Computer Sciences at the University of Wisconsin-Madison. My research focuses on ML Systems — making large models run faster and train more efficiently.

Current Work

I'm currently interning at AMD GenAI (Spring 2026), working on RL infrastructure and GPU kernel optimization. Specifically, I'm building multi-agent systems (MAS) to automatically optimize GPU kernels on MI300X / MI355X.

Research Interests

ML Systems: LLM inference optimization, serving frameworks (SGLang), GPU programming
RL Infrastructure: LLM post-training (SFT / RLHF / GRPO) system design and scaling
Kernel Optimization: Triton / CK kernel auto-generation and optimization, AMD ROCm ecosystem

Education

MS / PhD, Computer Sciences

University of Wisconsin-Madison · 2024 — present
BSE, Computer Science Engineering

University of Michigan · 2022 — 2024
BE, Electrical and Computer Engineering

Shanghai Jiao Tong University · 2020 — 2024

Open Source

I'm an active contributor to the SGLang community, and also contribute to slime, TritonForge, APRIL, and other projects.

See all projects →