News

• We released PromptCoT 2.0,  providing a new scaling knob-- prompt synthesis-- for large language model reasoning. 

• 5 papers accepted by NeurIPS 2025 (4 posters + 1 spotlight). The papers cover our recent progress on LLM reasoning (including a general approach and a special model for spatial reasoning), long-context modeling in LLMs,  RAG, and diffusion models. 

About Me

I am currently leading the NLP team at Ant Research, where my work focuses on advancing large language model (LLM) technologies, with particular emphasis on reasoning and agentic intelligence. Prior to this role, I served as Head of the NLP Team at Meituan (2020–2022), Principal Applied Scientist Lead in the Microsoft Xiaoice Team (2018–2020), and Lead Researcher at Microsoft Research Asia (MSRA) (2012–2017). Before joining MSRA as a full-time researcher, I was a joint Ph.D. student between MSRA and Peking University under the supervision of Professor Hang Li.

I received my B.S. in Applied Mathematics from Peking University in 2007 and earned my Ph.D. in Applied Mathematics from the same institution in 2012. I am also a Senior Member of the IEEE.

My research interests span over machine learning, natural language processing, and information retrieval, with a current focus on developing large language and vision-language models (LLMs and VLMs) capable of advanced reasoning across complex domains such as competition mathematics, programming, and spatial reasoning—and extending beyond these areas toward general intelligent systems.