Hello, I am Zhengye Han (韩政业), a second-year PhD student at New York University. I received my BS degree in Computer Science from Beijing Jiaotong University, where I studied under the supervision of Professor Zhigang Cao. Under his guidance, I embarked on my academic journey through the study of game theory.
During my undergraduate studies, I had the privilege to join the Center on Frontiers of Computing Studies (CFCS) at Peking University, working in the daGAME lab, where I focused on Reinforcement Learning and game theory. I am deeply grateful to Professor Xiaotie Deng for providing me with invaluable research and internship opportunities during this time. I would also like to extend my sincere thanks to Dr.Zhijian Duan for his important guidance throughout my time in the lab.
My research focuses on the following topics:
- Multi-Agent Systems: Understanding strategic interaction, coordination, and institutional mechanisms among intelligent agents, with a particular focus on diagnosing where and why multi-agent systems fail.
- Reinforcement Learning: Developing learning and planning methods for complex, interactive environments, with an emphasis on real-world applications such as large-scale auctions and autonomous driving.
Beyond AI, I draw inspiration from economics, physics, and biology.
Economics: I love economics because game theory and mechanism design offer a powerful lens for understanding how rules, incentives, and interactions shape the behavior of entire systems.
Physics and Biology: I often think of nature as the greatest designer, and I believe the natural world holds profound insights into how we can build intelligent collectives. I enjoy simulating the interactions and collisions of physical particles, as well as reading about how ant colonies coordinate and cooperate. To me, these are not just fascinating natural phenomena, but sources of inspiration for understanding, diagnosing, and designing multi-agent systems.
🔥 News
- Mar 2026 🎤 On Mar. 31, 2026, I was selected as a Top 8 finalist in NYU PhD Live!, becoming the only first-year Ph.D. student among the finalists. Video.
- Mar 2026 🎉 Our paper "Toward a Dynamic Stackelberg Game-Theoretic Framework for Agentic AI Defense Against LLM Jailbreaking" has been accepted by the ICLR 2026 Workshop on AI for Mechanism Design and Strategic Decision Making!
- Sep 2025 🗽 I officially started my Ph.D. journey at the ECE Department of New York University (NYU)!
- Jun 2024 🏆 Co-organized the NeurIPS 2024 Competition: "Auto-Bidding in Large-Scale Auctions" and open-sourced AuctionNet as a core code contributor.
Internship
For more details, please check my LinkedIn.
Selected Publications
- Zhengye Han. When Do Institutions Beat Intelligence? Under review at AAMAS 2027. [arXiv]
- Zhengye Han. Where Do Multi-Agent Systems Fail? Evidence-Grounded Diagnosis of Collective Mechanisms. Under review at AAMAS 2027 [arXiv]
- Zhengye Han. Safe Preference Identification through Strategic Interaction. Under review in 2027 International Conference on Learning Representations (ICLR).
- Zhengye Han, Quanyan Zhu. Toward a Dynamic Stackelberg Game-Theoretic Framework for Agentic AI Defense Against LLM Jailbreaking. Published in International Conference on Learning Representations (ICLR) AI for Mechanism Design Workshop, 2026. [arXiv] A substantially extended journal version is currently under review at IEEE Transactions on Information Forensics and Security (TIFS).
- Zhengye Han, Quanyan Zhu. RCSP: Risk-Sensitive Conjectural Scenario Planning for Safe Dynamic Robot Navigation. Under review in 2027 IEEE International Conference on Robotics & Automation (ICRA). [arxiv]
- Quanyan Zhu, Zhengye Han. Performative Scenario Optimization. Published in IEEE Conference on Decision and Control (CDC) 2026. [arXiv]
- Shuai Dou, Yusen Huo, Zhilin Zhang, Yeshu Li, Zhengye Han, Kefan Su, Zongqing Lu, Chuan Yu, Jian Xu, Bo Zheng. Code Contributor to AuctionNet: A Novel Benchmark for Decision-Making in Large-Scale Games. In NeurIPS 2024 Datasets and Benchmarks Track (Spotlight). [GitHub]
Project
InstitutionLab: Learning Collective Intelligence from Nature (in progress)
I think of nature as the greatest designer of collective intelligence. An ant colony has no central planner, yet it finds food, abandons exhausted sources and recovers when the world changes, because simple agents read and write shared traces in their environment. I believe the principles of such emergence are already written into the natural world, and studying them is how I want to understand, diagnose and design multi-agent systems.
InstitutionLab is the simulator I am building for this. Agents combine private memory, direct messages and a shared pheromone-like trail, while the rules that govern this shared state, such as how long traces last or what must be verified, are explicit, versioned and replayable. My first pre-registered experiments showed that agents can follow a shared trail heavily without gathering more food, so I now measure what a shared memory contributes, not how often it is used.
NeurIPS 2024 Competition Track: Auto-Bidding in Large-Scale Auctions
This competition was hosted by the PKU-Alimama Artificial Intelligence Innovation Joint Lab, a collaboration between Alibaba Group and Peking University. As a member of the lab, I was actively involved in organizing the competition. I worked closely with Shuai Dou and Yeshu Li to develop the agents for the competition system, focusing on creating robust models for decision-making in large-scale, competitive auctions.
Public welfare
Democratization of AI
I launched a public welfare campaign in Yulin City, Guangxi, to popularize AI knowledge, extending to high schools for public welfare lectures. The campaign primarily analyzed the impact of AI technology on the wealth gap and promoted the knowledge of using LLM.
Miscellaneous
- Public Speaking: NYU PhD live! top 8 PhD speaker, member of the School Hosting Team in BJTU.
- Singing: Campus Musician on NetEase Cloud Music, Participated in a choir performance at a national event on Mango TV.
- Hip-hop Dance: Champion of the Small Group Category in the Capital Colleges Dance Competition, Beijing.
- Writing: Passionate about literature, I served as the Deputy Director of the Original Editorial Department for the school’s official WeChat account, contributing to creative content and original articles.