about / blog / projects / ZH /

Jin Pan

ML Systems / LLM Inference / RL Infrastructure

Second-year MS/PhD student in Computer Sciences at UW-Madison, working on ML Systems. SGLang community contributor. Currently interning at AMD GenAI, focusing on RL systems and GPU kernel optimization.

More about me →

Recent Posts

Attention Mechanisms — Full, Sparse, Linear, NSA & GLA

Breaking down Full, Sparse, and Linear Attention, all the way to DeepSeek NSA and Gated Linear Attention
Benchmark: Qwen3-Coder-30B-A3B + EAGLE3 Speculative Decoding

EAGLE3 speculative decoding benchmarks on Qwen3-Coder — 1.87x speedup for code generation
NeMo-RL vs slime: RL Training Framework Comparison

Deep comparison of two RL training frameworks: algorithms, engineering quality, MoE support, and ROCm compatibility
TritonForge: Server-based Multi-turn RL for Triton Kernel Generation

End-to-end server-based RL training and evaluation system for Triton kernel generation across NVIDIA and AMD, built on slime + Megatron
SFT & RL Training Guide

Complete guide from SFT vs RL fundamentals, loss computation, dataset construction to RLHF in practice

Recent Projects

See all projects

Miles

Enterprise RL framework for LLM/VLM post-training. Integrates SGLang rollout + Megatron training with FP8 pipeline and MoE support.
SpecForge

Train speculative decoding draft models and port them to SGLang serving. Part of the SGLang ecosystem.
TritonForge

LLM-powered GPU kernel synthesis: Train models to convert PyTorch ops into optimized Triton kernels via SFT+RL.
APRIL

Active Partial Rollouts in Reinforcement Learning to Tame Long-tail Generation. A system-level optimization for scalable LLM training.
SGLang

High-performance serving framework for large language models and multimodal models. Contributor.

Contact

Find me on social media or send an email.

GitHub /
LinkedIn /
jpan236@wisc.edu

© 2026 • Jinn's Hub 🔬

Press Esc or click anywhere to close