About

Hello! I’m a PhD student at the University of Science and Technology of China (USTC), co-advised by Dr. Lei Bai and Prof. Wengang Zhou. My research interests are LLM Post-training, Agent and AI Infra. I earned my BSc in Computer Science from USTC, and was a Master’s student there before transferring to the PhD in Fall 2025. I am currently interning at Shanghai AI Lab, mentored by Dr. Chen Zhang and Dr. Shuyue Hu.

Feel free to reach out via email. You can also find me on Zhihu and Xiaohongshu.

My CV is available in English and 中文.

News

We released Sage-A1.5, a 35B MoE agentic model that advances long-horizon problem solving through adaptive capability composition, bringing together agentic orchestration, verification, and open scientific discovery.

Sage-A1.5 training pipeline, agentic orchestration, and verification overview

We released ScienceIDE, a platform that turns real scientific codebases, research workflows, task goals, and verification criteria into executable, verifiable, and reproducible environments for agent execution, model training, and reinforcement learning.

ScienceIDE environment, verification, SFT, and reinforcement-learning overview

We released SKT, a verified synthetic-data pipeline for training language-model agents to identify, apply, and coordinate reusable skills at scale.

SKT overview

We launched AgentPanel, a multi-agent forum for human-AI collaboration in exploring scientific questions. Accepted to DAI 2026 as an oral presentation.

AgentPanel interface

We released libsmctrl, a runtime library for SM-level GPU compute partitioning and operator isolation.

I joined Tencent as an AI Infra intern working on LLM inference acceleration, mentored by Hande Dong.