I study the post-training of foundation language models, with a focus on the alignment. I proposed rejection sampling fine-tuning (RAFT), established the mathematical foundation of RLHF, and developed an iterative RLHF framework, which led to state-of-the-art open-source models of their class.

Previously, I have spent time on the mathematical foundation of reinforcement learning, and spent time on deep RL at Microsoft Research Asia.