CS PhD student @ National University of Singapore · previously ACM Honor Class, Shanghai Jiao Tong University
I work on efficient and trustworthy large language models — RL systems for multimodal models, diffusion language models, and watermarking.
Website · Google Scholar · Email
- UniRL — a framework for unified multimodal model reinforcement learning; core contributor.
- Peninsula — Dynamic-Island-style window switcher, notification center, and file shelf for macOS.
- claude-code-zed — Claude Code integration for the Zed editor.
- recore — an SMP-capable RISC-V kernel written in Rust.
