Renmin University Turing grad Xiuyu Li joins StepFun as post-training researcher

Xiuyu Li, a graduate of Renmin University of China's prestigious Turing Class, has joined domestic LLM startup StepFun (阶跃星辰) as a post-training researcher.

During his time at university, Li focused on LLM alignment and reinforcement learning (RL). His academic contributions include serving as a co-first author on ETS, a training-free inference-time alignment method presented at ICML 2026, and DART, a paper on the conflict between reasoning and tool use in agent RL.

Note: This Xiuyu Li is not the same person as the former xAI member and Berkeley PhD Xiuyu Li.