I am a Research Intern at DeepWisdom, where I work with Chenglin Wu, Jiayi Zhang, and Yifan Wu. I am fortunate to collaborate with Bang Liu and Yuyu Luo.
I received my B.E. from the Renmin University of China. As a Co-Founder of OpenManus and a member of Foundation Agents, I am committed to advancing open-source agent infrastructure and research. Currently, my research interest focuses on developing LLM-based agents that can operate effectively across diverse environments and tasks.
Focus
- Agent learning. Learning is key to cross-environment capabilities. Learning environment dynamics requires complex optimization approaches, signals, and targets beyond model parameters, like AFlow optimizing decision workflows and SPO exploring new reward signals for prompt optimization.
- Decision-making. Human decision-making naturally enables cross-environment learning and generalization. We explore agent decision structures that mirror human reasoning, potentially unlocking similar learning advantages. AoT atomizes reasoning to address context limitations, while ReCode unifies planning and action for more natural decision-making.
- Environment scaling. Agent environments are inevitably simplified versions of human environments, lacking complexity, dynamics, and rich reward signals that make agent learning inherently challenging. We aim to scale environments to provide richer dynamics, diverse distributions, and more abundant rewards for effective learning.
I am actively seeking a PhD position.
Selected Publications
-
ReCode: Unify Plan and Action for Universal Granularity Control.
-
InteractComp: Evaluating Search Agents With Ambiguous Queries.
-
AFlow: Automating Agentic Workflow Generation.
-
Self-supervised Prompt Optimization.
-
Atom of Thoughts for Markov LLM Test-Time Scaling.
-
Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems.
-
VisJudge-Bench: Aesthetics and Quality Assessment of Visualizations.
Selected Projects
-
No fortress, purely open ground. OpenManus is Coming.
-
Next paradigm for LLM Agent. Unify plan and action through recursive code generation for adaptive, human-like decision-making.
-
About Awesome things towards foundation agents. Papers / Repos / Blogs / ...
-
🔥🔥🔥 ICLR 2025 Oral. Automating Agentic Workflow Generation.
Experience
-
Research Intern, DeepWisdom
Feb 2025 – Present · Shenzhen
with Chenglin Wu, Jiayi Zhang, and Yifan Wu -
Algorithm Engineer Intern, Xiaomi
Jan 2024 – Apr 2024 · Beijing
with MiRoboticsLab -
Research Assistant, GeWu Lab (Renmin Univ)
Jun 2022 – Aug 2023 · Beijing
with Prof. Di Hu
Service
- Reviewer: ICLR 2026; ICML 2025 MAS Workshop
Talks
-
“Advances and Challenges in Foundation Agents”
Invited talk at 2025 X-AGI & The 18th China-R Conference, Beijing (Oct 2025)
Blogs
-
规划与行动统一:ReCode 对 Agent 决策的重新思考 2025.11.03 · Chinese
用代码统一表示 plan 和 action,结合动态展开机制,可能为 foundation agent 的 learning 提供一个更好的 decision-making 基础,但这只是一个初步探索。
-
【小米CyberDog二次开发】让你的机器狗有自己的小情绪! 2022.09.14 · Chinese
基于小米CyberDog和文澜预训练模型,我们实现了铁蛋根据不同场景而做出不同的富有情绪动作的能力。下面是保姆级教程(基于Python实现),无需ROS相关知识,让你的铁蛋也拥有自己的情绪!