Shicheng Fan

Ph.D. Student in Computer Science · University of Illinois Chicago

prof_pic.jpg

First author papers: ICML 2026; NeurIPS 2026; COLM 2026

I am a Ph.D. student in Computer Science at the University of Illinois Chicago, advised by Prof. Philip S. Yu. I work closely with Prof. Kun Zhang at Carnegie Mellon University and have been a part-time Research Scientist Intern at Aether AI since August 2026.

I received my B.Eng. in Automation (Control & Robotics) from Zhejiang University’s Chu Kochen Honors College, where I worked with Prof. Qi Ye on dexterous robotic manipulation.

My research explores how AI agents can recursively improve their own memory, harnesses, and self-evolution mechanisms. I study how autonomous exploration and verifiable feedback help agents accumulate reusable knowledge and skills that guide subsequent exploration and task solving. Recent work includes RSIAgent for recursive exploration, CoEvoSkills for skill evolution, and Agentic Commerce World for evaluating agents in auditable environments.

I welcome research discussions and collaborations. Please feel free to reach out by email.

Experience

RSIAgent combines autonomous exploration and iterative memory refinement, using verified experience from each round to guide the next. Built on open-source GLM-5.3 and Kimi-K3, it surpasses GPT-6 Astra in reported partial-credit scores on OSWorld 2.0 (78.98 vs. 72.60) and Agents' Last Exam, Near-term (84.82 vs. 82.26).

Education

University of Illinois Chicago
Aug 2025 – Present

Ph.D. in Computer Science

Advised by Prof. Philip S. Yu · Research in Recursive Self-improvement of AI Agents and Causal Representation Learning

Zhejiang University
Sep 2021 – Jun 2025

B.Eng. in Automation (Control & Robotics)

Chu Kochen Honors College · GPA: 3.98/4.0 · Rank: 6/121 (Top 5%)

Outstanding Graduate · Dexterous robotic manipulation with Prof. Qi Ye

Invited Talks

RSIAgent
Invited talk at Kuaishou (Kling AI), Sep 2026
Agentic Commerce World
Invited talk at Visa, Sep 2026
CoEvoSkills
Invited talk at Alibaba, Jun 2026

Reviewer

2026 NeurIPS · ICLR · KDD · ACM Computing Surveys (CSUR)

news

Sep 2026
MOSAIC is accepted to NeurIPS 2026. See the code.
Sep 2026
We released RSIAgent, a framework for recursive self-improvement in new environments. Curriculum, actor, and verifier agents autonomously explore, verify outcomes, and consolidate experience into memory that guides subsequent rounds. Built on open-source GLM-5.3 and Kimi-K3, it outperforms GPT-6 Astra in partial-credit scores on OSWorld 2.0 and Agents’ Last Exam. Hugging Face Daily Papers #6 (Sep 15, 2026); Headline feature in Synced (机器之心), an AI news outlet. See the code and project page.
Aug 2026
I started a part-time research internship at Aether AI.
Aug 2026
We released Agentic Commerce World (ACWorld), an auditable and verifiable environment for agentic commerce. See the code.
Jul 2026
We released Paying for Honesty Without Knowing the Truth, on reputation-penalty design for LLM marketplace agents.
Jul 2026
CoEvoSkills is accepted to COLM 2026.
Jun 2026
We released Scaling LLM Agent Learning with Data Synthesis, a comprehensive survey of data synthesis for LLM agents.
May 2026
We released CorVer, lightweight corpus-grounded process supervision for factual question answering. See the code.
May 2026
We released MOSAIC, a method for interpretable module discovery in scientific time series. See the code.
Apr 2026
TRACE is accepted to ICML 2026.
Apr 2026
We released CoEvoSkills, self-evolving agent skills via co-evolutionary verification. See the code.
Jan 2026
We released TRACE, trajectory recovery for continuous mechanism evolution in causal representation learning. See the code.
Aug 2025
I started my Ph.D. in Computer Science at the University of Illinois Chicago.

selected publications

  1. Preprint
    RSIAgent.png
    RSIAgent: Autonomous Exploration for Recursive Self-improvement in New Environments
    Sibo Zhu*, Shicheng Fan*, Xinyue Wang*, Wenyi Wu, Kun Zhou, and Biwei Huang
  2. Preprint
    ACWorld.png
    Agentic Commerce World: An Auditable and Verifiable Environment for Vibe Commerce
    Shicheng Fan, Mingdai Yang, Duohao Wang, Canyu Chen, Yongfeng Zhang, Hua Wei, and 6 more authors
    2026
  3. COLM 2026
    CoEvoSkills.png
    CoEvoSkills: Self-Evolving Agent Skills via Co-Evolutionary Verification
    Hanrong Zhang*, Shicheng Fan*, Henry Peng Zou, Yankai Chen, Zhenting Wang, Jiayu Zhou, and 7 more authors
    In Conference on Language Modeling (COLM), 2026
  4. ICML 2026
    Trace.png
    TRACE: Trajectory Recovery for Continuous Mechanism Evolution in Causal Representation Learning
    Shicheng Fan, Kun Zhang, and Lu Cheng
    In International Conference on Machine Learning (ICML), 2026
  5. NeurIPS 2026
    MOSAIC-combined.png
    MOSAIC: Module Discovery via Sparse Additive Identifiable Causal Learning for Scientific Time Series
    Shicheng Fan, Nour Elhendawy, Jianle Sun, Ke Fang, Kun Zhang, Yihang Wang, and 1 more author
    In Conference on Neural Information Processing Systems (NeurIPS), 2026
  6. Preprint
    CorVer.png
    Beyond Math and Code: Lightweight Corpus-Grounded Process Rewards for Factual Question Answering
    Shicheng Fan*, Haochang Hao*, Dehai Min*, Weihao Liu, Hanrong Zhang, Lingwei Wei, and 7 more authors
    Under review at ICLR 2027 , 2026