Mingchen Zhuge

*Taken in the KAUST AI Initiative.

Mingchen Zhuge

I am a PhD candidate with a strong interest in multimodal agents and world models, including their training (especially post-training) and applications. I am fortunate to have Prof. Jürgen Schmidhuber as my advisor. I have 20+ top-tier publications and over 3,300 citations (80% from first-author papers). I enjoy contributing to open-source frameworks such as MetaGPT GitHub Stars, GPTSwarm GitHub Stars, OpenDevin (aka. OpenHands) GitHub Stars, and agent-as-a-judge GitHub Stars. My papers on multimodal agents were accepted as oral presentations at ICLR (<1.2%), ICML (<1.5%) or received the Best Paper Award at the NeurIPS Ro-FoMo Workshop. I was nominated for WAIC Future Star Nomination (明日之星提名奖, 15人, 2025) and recognized as an Outstanding Reviewer of CVPR2023 (232 out of 7000+).

📣 I am on the 2025–2026 job market and wanna find a place with a strong ambition for future superintelligence.

Twitter GitHub Google Scholar LinkedIn
KAUST Meta Microsoft Alibaba IIAI

Selected Publications ( Full list)

2024

Agent-as-a-Judge Image
Agent-as-a-Judge: Evaluate Agents with Agents
Mingchen Zhuge, Changsheng Zhao, Dylan Ashley, Wenyi Wang, Dmitrii Khizbullin, Yunyang Xiong, Zechun Liu, Ernie Chang, Raghuraman Krishnamoorthi, Yuandong Tian, Yangyang Shi, Vikas Chandra, Jürgen Schmidhuber
ICML 2025
The first paper introduce the philosophy of Agent-as-a-Judge.
Dramatically reduces evaluation costs while maintaining high correlation with human judgments.
[Paper] [Code] GitHub Stars [BibTex] [Tweet1] [Tweet2] [机器之心] [AI Era]
Image Description
OpenHands: An Open Platform for AI Software Developers as Generalist Agents
OpenHands Open-sourced Community (Lead general agent in the paper)
ICLR 2025
Achieved 50k+ stars on GitHub.
A comprehensive single-agent platform for AI software developers as generalist agents.
[Paper] [Code] GitHub Stars [BibTex] [机器之心]
Image Description
GPTSwarm: Language Agents as Optimizable Graphs
Mingchen Zhuge, Wenyi Wang, Louis Kirsch, Francesco Faccio, Dmitrii Khizbullin and Jürgen Schmidhuber
ICML 2024 (Oral)
Oral Presentation (top 1.5% in 9,473).
First framework using graph to build LLM agents, enabling node and edge optimization.
[Paper] [Code] GitHub Stars [BibTex] [MIT Tech Review Interview] [Jiangmen Ventures]

2023

Image Description
MetaGPT: Meta Programming For A Multi-Agent Collaborative Framework
Sirui Hong∗, Mingchen Zhuge∗, Jonathan Chen, Xiawu Zheng, Yuheng Cheng, Ceyao Zhang, Jinlin Wang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, Chenyu Ran, Lingfeng Xiao, Chenglin Wu†, Jürgen Schmidhuber
ICLR 2024 (Oral)
Oral Presentation (top 1.2% in 7,262).
Achieved 50k+ stars on GitHub, pioneering multi-agent collaborative software development.
[Paper] [Code] GitHub Stars [BibTex] [QianTang Report]
Image Description
Mindstorms in Natural Language-Based Societies of Mind
Mingchen Zhuge*, Haozhe Liu*, Francesco Faccio*, Dylan R. Ashley*, Róbert Csordás, Anand Gopalakrishnan, Abdullah Hamdi, Hasan Abed Al Kader Hammoud, Vincent Herrmann, Kazuki Irie, Louis Kirsch, Bing Li, Guohao Li, Shuming Liu, Jinjie Mai, Piotr Piękos, Aditya Ramesh, Imanol Schlag, Weimin Shi, Aleksandar Stanić, Wenyi Wang, Yuhui Wang, Mengmeng Xu, Deng-Ping Fan, Bernard Ghanem, Jü̈rgen Schmidhuber
NeurIPS Ro-FoMo 2023 (Oral, Best Paper Award); CVM Journal
Best Paper Award at NeurIPS Ro-FoMo Workshop.
First Position Paper introducing the agentic society (society of mind) and agentic economy (economy of mind).
[Paper] [Code] GitHub Stars [BibTex] [Poster] [Award]

2022

Image Description
Salient Object Detection via Integrity Learning
Mingchen Zhuge*, Deng-Ping Fan*, Nian Liu, Dingwen Zhang, Dong Xu, Ling Shao
TPAMI 2022
ESI Highly Cited Paper.
One of earliest open-sourced codebases that utilize Transformer/MLP encoder for foreground segmentation.
[Paper] [Code] GitHub Stars [BibTex]

2021

Image Description
Kaleido-BERT: Vision-Language Pre-training on Fashion Domain
Mingchen Zhuge*, Dehong Gao*, Deng-Ping Fan†, Linbo Jin, Ben Chen, Haoming Zhou, Minghui Qiu, Ling Shao
CVPR 2021
Industry Application in Alibaba.com.
The second VLP model in e-commerce, introducing dynamic-resolution self-supervised representation learning, one of 4 VLP papers at CVPR 2021 (not many researchers work on multimodal learning at that time).
[Paper] [Code] GitHub Stars [BibTex] [Talk]

Articles/News

MIT Technology Review
Jürgen Schmidhuber's team presents new research: Using graph structures to build agents, advancing AI agent development
March 13, 2024
Read Article
Synced (机器之心)
Raising the bar! Evaluating agents with agents, Meta releases Agent-as-a-Judge
October 18, 2024
Read Article
AI Era (新智元)
Yuandong Tian's team introduces Agent-as-a-Judge! AI agents self-evaluation reduces costs by 97%
October 27, 2024
Read Article

Selected Awards

2025
WAIC 明日之星提名 (15人)
2023
Best Paper Award in NeurIPS Ro-FoMo Workshop
2023
Outstanding Reviewer of CVPR2023 (232 out of 7000+)
2021
2021 ZTE Blue Sword Program (Only three selected from Wuhan universities)
2019
16th China Graduate Mathematical Modeling Competition - National Second Prize
2019
2019 ZTE Algorithm Competition Champion (Top 7/11000+)
© 2024 Mingchen Zhuge counter
GitHubGitHub TwitterTwitter Google ScholarGoogle Scholar EmailContact