Baidu
Foundation Model Algorithm Intern, ERNIE Team · May 2026 – Present
Selected for the ERNIE Rising Star Top-Talent Program.
Agentic reinforcement learning for coding on the ERNIE Lite model.
- Reward Modeling and Optimization Objective Design
- Credit Assignment for Long-Horizon Agentic Trajectories
- Reward Hacking Governance and Trajectory Attribution
- Single-Step Distillation Training at Production Scale