Hi there! I’m currently a postdoctoral researcher at the University of Oxford, supervised by Prof. Philip Torr. I’m also visiting Le Cong and Mengdi Wang’s Lab at Stanford and Princeton. I did my Ph.D. at the University of Sydney, working with Prof. Wanli Ouyang and Prof. Zhiyong Wang. Previously, I was a rising star research fellow at the Shanghai AI Lab selected by Prof. Xiaoou Tang, where I collaborated with outstanding researchers like Dr. Lei Bai, and Dr. Amanda Shao. I also had a wonderful time as a visitor at the Chinese University of Hong Kong. Before starting my Ph.D., I was part of SenseTime’s AGI group, working closely with Dr. Junjie Yan. I earned my bachelor’s degree from HUST, where I had the honor of being the ACM-ICPC team captain, guided by Prof. Kun He.

News

I’m on the academic job market in 2026. Curriculum Vitae
To junior students seeking advice on early academic careers: If you’d like to chat about your career, research ideas, or potential collaborations, feel free to email me to schedule a meeting. I’d be happy to recommend some internship or study opportunities as well.
I’m looking for motivated students to work with me on topics such as agentic AI, multi-agent systems, embodied agents, post-training of multi-modal large language models, and related areas. A strong background in these fields is a plus, but not a must; curiosity, commitment, and a willingness to learn matter most. If you’re interested, please email me with your CV and a short note about your research interests.
2026.06: I am organizing four CVPR 2026 workshops: 2nd Workshop on Multi-Modal Reasoning for Agentic Intelligence (MMRAgI), ScaleBot: The First Workshop on Scalable Robot Learning Systems, Agentic AI for Visual Media, and the 6th Workshop on Adversarial Machine Learning on Computer Vision: Safety of Vision-Language Agents.
2026.04: I am organizing two ICLR 2026 workshops: Lifelong Agents: Learning, Aligning, Evolving and the First Workshop on Efficient Spatial Reasoning.
2026.04: I will give an invited talk on “Agents and World Models in Wet Lab” at the ICLR 2026 workshop on World Models: Understanding, Modelling, and Scaling.
2026.04: I will participate in a panel discussion at the ICLR 2026 workshop on AI with Recursive Self-Improvement.
2025.10: We are organizing the SFE Challenge, focusing on advancing the capabilities of foundation models as AI Scientist base models. The competition has just begun, and everyone is warmly invited to participate and contribute innovative ideas!
2025.08: We are organizing a competition on multi-agent embodied intelligence. All relevant details, datasets, and the platform have been released; please visit the official website for further information.
2025.10: I organized the ICCV 2025 workshop on Reliable and Interactable World Models: Geometry, Physics, Interactivity and Real-World Generalization.
2025.10: I organized the ICCV 2025 workshop on Multi-Modal Reasoning for Agentic Intelligence.
2025.09: Thrilled to release our survey + position paper on LLM Agent Reinforcement Learning, where we systematically define and outline the emerging paradigms of LLM RL and LLM Agent RL. Check it out on Hugging Face, and explore our curated Awesome paper list. If you find it helpful, please consider an upvote or a star!
2025.07: Our work VirSci was featured in Nature Feature [PDF]! The concept of Co-AI Scientists is gaining wide attention.
2025.07: Our paper “Revealing the Risks of Utilizing Large Language Models in Scholarly Peer Review” was covered by The Washington Post [PDF]. The piece discusses the potential impact of LLM-based reviewing on the future of the peer review ecosystem.
2025.07: Honored to be selected as a WAIC 2025 Yunfan Award Rising Star nominee, recognizing emerging researchers in the field of AI.
2025.07: I organized the ICML 2025 workshop on Multi-Agent Systems in the Era of Foundation Models: Opportunities, Challenges, and Futures (MAS-2025), which was one of the most well-attended events of the entire conference!
2025.04: Thrilled to release MARS (Multi-Agent Robotics System), an open-source framework focusing on embodied intelligence in multi-agent settings. MARS aims to support almost all approaches based on foundation model embodied agents, spatial intelligence, and compositional intelligence (generalization and constraints). You’re welcome to follow and contribute!
2025.04: Excited to announce MASWorks/MASLab (a nod to MathWorks/Matlab!), an open-source framework dedicated to multi-agent systems based on LLM agents, providing all essential components for MAS research, datasets, benchmarks, codebases, and more. We’ll also be releasing a series of new research projects based on this platform. Join us in building the community!
2024.12: I gave a talk at the NeurIPS 2024 Workshop on Open-World Agents, titled “Building AI Society with Foundation-Model Agents.”
2024.11: Thrilled to announce OASIS, a simulation platform supporting interactions among over one million LLM agents.
2024.07: I organized the ICML 2024 workshop on Multi-modal Foundation Models Meet Embodied AI (MFM-EAI).
2024.07: I organized the ICML 2024 workshop on Trustworthy Multi-modal Foundation Models and AI Agents (TiFA).
2024.05: I co-hosted the EgoPlan Challenge to evaluate embodied agents’ complex planning capabilities.
2023.11: Excited to release LAMM, a comprehensive framework for VLM training, evaluation, and applications in embodied agents.
2023.08: I began organizing a weekly academic talk series, Echo AI Talk, inviting young researchers from around the world who are well-known for their work in generative AI, foundation models, and AI agents. Everyone is welcome to join!
2021.11: Excited to release Intern, a series of multi-modal foundation models focusing on visual representation learning.
2020.07: Achieved Rank 4 of 2265 in Meta’s DFDC competition, which focused on identifying videos with facial or voice manipulations. Our solution is open-sourced.
2018.05: As a student coach, I led a team to the ACM-ICPC World Finals, achieving 31st place.

Research Highlights

My research aims to develop general-purpose AI agents capable of operating in both physical and virtual environments. My work primarily focuses on foundation model post-training, multi-agent systems, self-evolving agents, and embodied agents and robotics, with the goal of improving the autonomy, adaptability, and coordination capabilities of intelligent systems in complex environments.

A central application of this research is the development of a General AI Scientist for scientific discovery across both natural sciences and social sciences. These systems leverage AI agents and embodied robotic agents to autonomously explore research problems, generate hypotheses, conduct simulations or experiments, and discover new knowledge. Through this framework, my research aims to uncover new scaling laws of agent-based intelligence and automated scientific discovery in large-scale interactive environments.

Selected Publications

Topics: Foundation Model Agents / Robotics / AI Scientists

(*: indicates equal contribution; ‡: indicates corresponding; †: indicates project lead)

Preprint

Memory in the Age of AI Agents

Yuyang Hu^*, Shichun Liu^*, Yanwei Yue^*, Guibin Zhang^*, Boyang Liu, Fangyi Zhu, Jiahang Lin, Honglin Guo, Shihan Dou, Zhiheng Xi, Senjie Jin, Jiejun Tan, Yanbin Yin, Jiongnan Liu, Zeyu Zhang, Zhongxiang Sun, Yutao Zhu, Hao Sun, Boci Peng, Zhenrong Cheng, Xuanbo Fan, Jiaxin Guo, Xinlei Yu, Zhenhong Zhou, Zewen Hu, Jiahao Huo, Junhao Wang, Yuwei Niu, Yu Wang, Zhenfei Yin, Xiaobin Hu, Yue Liao, Qiankun Li, Kun Wang, Wangchunshu Zhou, Yixin Liu, Dawei Cheng, Qi Zhang, Tao Gui, Shirui Pan, Yan Zhang, Philip Torr, Zhicheng Dou, Ji-Rong Wen, Xuanjing Huang, Yu-Gang Jiang, Shuicheng Yan

Preprint 2025

Zhenfei (Jeremy) Yin

News

Research Highlights

Selected Publications

Professional Service