I am Yuhan Chi (池裕涵), pronounced /juːˈhɑːn tʃiː/, an undergraduate at the School of Mathematical Sciences, Fudan University.
I enjoy distilling complex engineering problems into elegant mathematical models and thinking about how to make decisions under uncertainty. My current research focuses on the intersection of reinforcement learning and generative models.
I also care deeply about socioeconomic issues and contribute to open-source projects for learning, sharing, and community life.
Feel free to reach out! I’m always happy to connect over coffee, sports, or a shared interest.
Optimization, control, and reinforcement learning for sequential decision-making and generative models; mathematical modeling for real-world problems.
Studied hardware interfaces and explored dexterous-hand grasping in simulation.
What can behavior identify—and when is that enough for reliable decisions?
Adapted conformal inverse optimization to IRL using a coordinate-invariant score based on policy values. Quantified the assumptions needed to turn behavioral compatibility into true-reward coverage, and proved when the resulting robust planner cannot certify improvement over a reference policy.
Can cheap reasoning-trace features predict correctness across math, science, and code?
Using these features to select among 64 candidates changes pass@1 by +10.0 pp in math, +8.0 pp in science, and −0.6 pp in code.
Can a tiny shared parameterization learn code generation from executable feedback?
A GRPO training pipeline for Qwen2.5-Coder-3B that rewards real compile-and-run outcomes instead of static heuristics.
From Bandits to PPO: RL Prerequisite Seminar — Fudan University, Apr 2026 (PDF)