Welcome!

This is Peixuan Han, a second-year PhD student at the University of Illinois, Urbana-Champaign, School of Computing and Data Science, advised by Prof. Jiaxuan You. Before coming to UIUC, I earned my bachelor’s degree at Tsinghua University, Department of Computer Science. I’ve also interned at Carnegie Mellon University in 2023 and at Amazon AWS in 2025.

My research focuses on Natural Language Processing. I strive to develop language agents that can adapt to complex environments and solve real-world issues efficiently and safely. Specifically, my current research focuses are: 1) goal-oriented dialogs, including persuasion and academic rebuttal; 2) social reasoning, social intelligence, and safety of LLMs.

To learn more about me, check my CV here! I’m actively looking for research internships in Summer 2026, feel free to contact me if you think there may be a match.

Recent Publication

• Peixuan Han, Adit Krishnan, Gerald Friedland, Jiaxuan You, Chris Kong. Self-Aligned Reward: Towards Effective and Efficient Reasoners

• Peixuan Han, Zijia Liu, Jiaxuan You. ToMAP: Training Opponent-Aware LLM Persuaders with Theory of Mind

• Peixuan Han, Cheng Qian, Xiusi Chen, Yuji Zhang, Denghui Zhang, Heng Ji. Internal Activation as the Polar Star for Steering Unsafe LLM Behavior (EMNLP’25)

• Tao Feng, Haozhen Zhang, Peixuan Han, Zijie Lei, Jiaxuan You. GraphPlanner: Graph-Based Agentic Routing for LLMs

• Hongyi Du, Jiaqi Su, Jisen Li, Lijie Ding, Yingxuan Yang, Peixuan Han, Xiangru Tang, Kunlun Zhu, Jiaxuan You. Which LLM Multi-Agent Protocol to Choose?

• Xiusi Chen, Shanyong Wang, Cheng Qian, Hongru Wang, Peixuan Han, Heng Ji. DecisionFlow: Advancing Large Language Model as Principled Decision Maker (EMNLP’25)

• Kunlun Zhu, Jiaxun Zhang, Ziheng Qi, Nuoxing Shang, Zijia Liu, Peixuan Han, Yue Su, Haofei Yu, Jiaxuan You. SafeScientist: Toward Risk-Aware Scientific Discoveries by LLM Agents (EMNLP’25)

• Alexi Gladstone, Ganesh Nanduru, Md Mofijul Islam, Peixuan Han, Hyeonjeong Ha, Aman Chadha, Yilun Du, Heng Ji, Jundong Li, Tariq Iqbal. Energy-Based Transformers are Scalable Learners and Thinkers

• Zijia Liu, Peixuan Han, Haofei Yu, Haoru Li, Jiaxuan You. Time-R1: Towards Comprehensive Temporal Reasoning in LLMs