people

I am a PhD student at Tongji University, advised by Prof. Gang Yan, and a visiting researcher at the University of Illinois Urbana-Champaign (UIUC), advised by Prof. Heng Ji, Prof. Dilek Hakkani-Tur, and Prof. Jiaxuan You. My research is centered on pushing the boundaries of what Large Language Models (LLMs) can achieve, with a focus on building more capable, general, and responsible AI agents through advanced Reinforcement Learning.

My recent achievement is leading the Time-R1 project, where my novel RL framework empowered a 3B model to outperform models 200x larger on challenging temporal reasoning benchmarks. This work underscores my expertise in creating highly efficient training pipelines that prioritize sophistication over sheer scale.

My experience also includes: • Architecting general-purpose agent frameworks like OpenManus to enhance strategic planning and generalization. • Integrating robust safety and ethical layers into AI scientist agents through the SafeScientist framework. • Designing persuasive agents with Theory of Mind to improve their interaction capabilities.

Core Competencies: • Agentic AI: Reasoning, Planning, Memory, Generalization • Reinforcement Learning: Online/Offline RL, Reward Engineering, Curriculum Learning • LLM Specializations: Multimodality (LVLM), Temporal Reasoning, Persuasive AI, AI Safety

I am passionate about engaging with fellow researchers and practitioners. I am always open to discussing new research, challenges, and potential collaborations!

I am a PhD student at Tongji University, advised by Prof. Gang Yan, and a visiting researcher at the University of Illinois Urbana-Champaign (UIUC), advised by Prof. Heng Ji, Prof. Dilek Hakkani-Tur, and Prof. Jiaxuan You. My research is centered on pushing the boundaries of what Large Language Models (LLMs) can achieve, with a focus on building more capable, general, and responsible AI agents through advanced Reinforcement Learning.

My recent achievement is leading the Time-R1 project, where my novel RL framework empowered a 3B model to outperform models 200x larger on challenging temporal reasoning benchmarks. This work underscores my expertise in creating highly efficient training pipelines that prioritize sophistication over sheer scale.

My experience also includes: • Architecting general-purpose agent frameworks like OpenManus to enhance strategic planning and generalization. • Integrating robust safety and ethical layers into AI scientist agents through the SafeScientist framework. • Designing persuasive agents with Theory of Mind to improve their interaction capabilities.

Core Competencies: • Agentic AI: Reasoning, Planning, Memory, Generalization • Reinforcement Learning: Online/Offline RL, Reward Engineering, Curriculum Learning • LLM Specializations: Multimodality (LVLM), Temporal Reasoning, Persuasive AI, AI Safety

I am passionate about engaging with fellow researchers and practitioners. I am always open to discussing new research, challenges, and potential collaborations!