Yuejiang Liu

Postdoc, Department of Computer Science, Stanford University
Assistant Professor (Incoming), National University of Singapore

Email · Scholar · GitHub · Twitter · RedNote · Group

Hi, there! I am currently a postdoctoral fellow at Stanford University, where I am fortunate to be advised by Chelsea Finn and mentored by Yilun Du. Previously, I obtained my PhD from EPFL, where I was fortunate to be advised by Alexandre Alahi, and interned with Francesco Locatello, Chris Russell, and Bernhard Schölkopf.

I will be joining the Department of Computer Science at the National University of Singapore (NUS) as an Assistant Professor, supported by the Presidential Young Professorship award. We are actively looking for PhD students and research assistants to join our LEMA group starting in 2027. If you are interested in joining us or exploring potential collaborations, we would be happy to hear from you.

Research

My research aims to build intelligent agents that can perceive, reason, and act in the physical world. I am particularly interested in physical intelligence in open dynamic worlds, where robot data is fundamentally limited, while moving objects, partial observability, and other interactive agents are the norm. To address these challenges, my research spans three interconnected elements:

Adaptive robot policies

How can robot policies adapt dynamically at inference time to new tasks, environments, and embodiment variations?

Interactive world models

How can structured world models help agents anticipate spatial, temporal, and physical interactions before acting?

Proactive self-improvement

How can agents verify, curate, and explore informative experiences to continually improve with minimal human supervision?

Our recent research has been recognized with paper awards at workshops at ICLR, CVPR, and RSS, alongside invited talks at OpenAI, NVIDIA, and other institutions..

Recent Publications

* Equal contribution. † Equal advising. For the full list, please refer to my Google Scholar.

World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetry

Yuejiang Liu*, Fan Feng*, Lingjing Kong*, Weifeng Lu*, Jinzhou Tang, Kun Zhang, Kevin Murphy, Chelsea Finn, Yilun Du

Preprint, 2026. Best Paper Award, ICLR World Model Workshop.

Paper · Project · Code

Scaling Verification Can Be More Effective than Scaling Policy Learning for Vision-Language-Action Alignment

Jacky Kwok, Xilun Zhang, Mengdi Xu, Yuejiang Liu†, Azalia Mirhoseini†, Chelsea Finn†, Marco Pavone†

Preprint, 2026. Best Paper Finalist, CVPR Scalable Robot Learning Workshop.

Paper · Project · Code

RoboMME: Benchmarking and Understanding Memory for Robotic Generalist Policies

Yinpei Dai, Hongze Fu, Jayjun Lee, Yuejiang Liu, Haoran Zhang, Jianing Yang, Chelsea Finn, Nima Fazeli, Joyce Chai

International Conference on Machine Learning (ICML), 2026. Oral (0.7%).

Paper · Project · Code

Learning Long-Context Diffusion Policies via Past-Token Prediction

Marcel Villasevil*, Andy Tang*, Yuejiang Liu*, Chelsea Finn

Conference on Robot Learning (CoRL), 2025. Best Paper Award, RSS Robot Representation Workshop.

Paper · Project · Code

Bidirectional Decoding: Improving Action Chunking via Guided Test-Time Sampling

Yuejiang Liu*, Jubayer Ibn Hamid*, Annie Xie, Yoonho Lee, Max Du, Chelsea Finn

International Conference on Learning Representations (ICLR), 2025. Invited talk at OpenAI.

Paper · Project · Code

Contact

Research

For research discussions and academic service, please contact me at yuejiang.liu [at] cs.stanford.edu.

Openings

For PhD / RA applications, please review openings in my incoming group and submit the application form. If you do not hear back within one week and believe there is a strong fit, you may follow up at liuyj [at] comp.nus.edu.sg.