Chi works on LLM post training infrastructures at Bytedance Seed. He focuses on optimizing RL framework training throughout, as well as algorithm infrastructure co-design for RL workloads including reasoning and agentic models. Chi initiates and prototypes the verl project at Bytedance, a LLM training framework with RL. Before joining Bytedance, he received his PhD at USC and worked with professor Viktor Prasanna in building distributed reinforcement learning systems.
Projects
verl: a flexible, efficient and production-ready RL training library for large language models (LLMs).
Seed-Thinking-v1.5: Advancing Superb Reasoning Models with Reinforcement Learning
Papers
HybridFlow: A Flexible and Efficient RLHF Framework, Eurosys 2025