Yafu Li

https://yafuly.github.io/I work on Large Language Models (LLMs), focusing on reasoning, reinforcement learning, and agent systems. My research aims to improve how models think—making them more accurate, efficient, and controllable under realistic compute constraints. My recent work focuses on: - Efficient reasoning and inference-time scaling - Reinforcement learning for large reasoning models - Proactive agent systems (e.g., OpenClaw) I have published at top-tier venues including NeurIPS, ICML, ICLR, ACL, and EMNLP. My work aims to bridge theoretical insights with practical system design, particularly for scalable and efficient AI. Previously, I worked on production-grade machine translation systems in industry. I remain interested in deploying large-scale models efficiently in real-world settings. I am open to collaborations on LLM reasoning, reinforcement learning, and scalable AI systems.