Ying Liu

Senior Director of Research

Scale AI

San Francisco Bay Area

I am currently Senior Director of Research at Scale AI, where I lead Agents Research.

On the product side, my team builds RL environments for agents at scale: turning real codebases, documents, web apps, and enterprise workflows into verifiable, hard-to-exploit tasks that serve as both public benchmarks and RL post-training data.

On the research side, we work on a broad range of topics: agent capabilities and safety (e.g., coding, computer use, tool use), professional-domain agents, agent training foundation, agentic workflows for producing frontier data, and efficient multi-agent systems. I am particularly interested in AI for Science, and work with universities and research institutes to design and build agentic laboratories and automated research systems.

Before Scale, I spent seven years at Meta, most recently building its large-scale RL platform for LLM post-training and video generation, and earlier leading teams working on recommendation systems. Prior to that, I spent five years at Google on hotel search monetization and web search ranking. I hold a Ph.D. in EECS from MIT and a B.E. from Tsinghua University.

See research and publications for more.

news

Aug 2026	I co-authored a perspective on Agentic Laboratories of the Future: Towards World Models for Scientific Discovery, a joint effort spanning 19 institutions, including Princeton University, Stanford University Columbia University, NVIDIA, Massachusetts Institute of Technology, Harvard University, Caltech, and Scale AI.
Jul 2026	Two talks at ICML 2026 in Seoul: an expo talk on Frontiers in Evaluation, Rewards, and Agent Environments, and Towards Autonomous and Self-Improving Agents at the 2nd Workshop on Agents in the Wild.
Jun 2026	Joined Scale AI as Senior Director of Research, leading Agents Research.

latest posts

selected publications

arXiv

Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-as-Reward RL

Minglai Yang, Xinyu Guo, Utkarsh Tyagi, Mian Zhang, Razvan Dumitru, Sunjie Hou, Yunzhong He, Daniel Yue Zhang, and Ying Liu

arXiv preprint arXiv:2608.11669, Aug 2026

arXiv Bib

@article{yang2026rubric,
  title = {Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-as-Reward {RL}},
  author = {Yang, Minglai and Guo, Xinyu and Tyagi, Utkarsh and Zhang, Mian and Dumitru, Razvan and Hou, Sunjie and He, Yunzhong and Zhang, Daniel Yue and Liu, Ying},
  journal = {arXiv preprint arXiv:2608.11669},
  year = {2026},
  month = aug,
}

Preprint

Agentic Laboratories of the Future: Towards World Models for Scientific Discovery

Kelsey Fu, Luna Lyu, Shuzhen Li, Suozhi Huang, Suheng Xu, Joy Zheng, Xuefeng Liu, Shilong Liu, Greg Barbone, Ying Liu, Linxi Jim Fan, Yuke Zhu, Matthew Davis, Kaiyi Jiang, Sherry Yang, and 40 more authors

Preprints.org, Aug 2026

DOI Bib HTML PDF

@article{fu2026agentic,
  title = {Agentic Laboratories of the Future: Towards World Models for Scientific Discovery},
  author = {Fu, Kelsey and Lyu, Luna and Li, Shuzhen and Huang, Suozhi and Xu, Suheng and Zheng, Joy and Liu, Xuefeng and Liu, Shilong and Barbone, Greg and Liu, Ying and Fan, Linxi Jim and Zhu, Yuke and Davis, Matthew and Jiang, Kaiyi and Yang, Sherry and Kuang, Xu and Luo, Yuan and Dominici, Francesca and Curtis, Christina and Qiu, Xiaojie and Abadie, Alberto and Uhler, Caroline and Tang, Mary and Wu, Joseph C. and Li, Ju and Toettcher, Jared and Avalos, Jos{\'e} L. and Rojansky, Rebecca and Zhao, Huimin and Ioannidis, John P. A. and Rand, Barry and Lundberg, Emma and Rosen, Andrew S. and Liu, Zhuang and Kelley, Shana and Xiong, Chenyan and Engelhardt, Barbara E. and Montine, Thomas and Theodoris, Christina V. and Pollard, Katherine S. and Li, Fuxin and Zhang, Xueyue and Peng, Tianyi and Basov, Dmitri and Qian, Xiang and Xing, Eric and Yazdani, Ali and Leskovec, Jure and Shah, Nigam and Bao, Zhenan and Perona, Pietro and Wu, Sanfeng and Brangwynne, Clifford and Cong, Le and Wang, Mengdi},
  journal = {Preprints.org},
  year = {2026},
  month = aug,
  doi = {10.20944/preprints202608.0213.v1},
}

Report

DrugDiscoveryBench: Can Coding Agents Assist Early-Stage Drug Discovery?

Afra Feyza Akyürek, Xinming Tu, Alec Gutmanstein, Jason Qin, Divyansh Agarwal, Sofia Monasdotter, Sergey Chekhov, Brenda Hernandez Villegas, Kirill Chugunov, Judah Engel, Veronica Chatrath, Oscar Kavanagh, Geobio Boo, Ernesto Hernandez, Ying Liu, and 7 more authors

Scale AI technical report, with Phylo , Jun 2026

Bib HTML PDF

@techreport{akyurek2026drugdiscoverybench,
  title = {{DrugDiscoveryBench}: Can Coding Agents Assist Early-Stage Drug Discovery?},
  author = {Aky{\"u}rek, Afra Feyza and Tu, Xinming and Gutmanstein, Alec and Qin, Jason and Agarwal, Divyansh and Monasdotter, Sofia and Chekhov, Sergey and Hernandez Villegas, Brenda and Chugunov, Kirill and Engel, Judah and Chatrath, Veronica and Kavanagh, Oscar and Boo, Geobio and Hernandez, Ernesto and Liu, Ying and Xue, Yuan and Sabharwal, Aakash and Zhang, Daniel Yue and Doctor, Zainab and Qu, Yuanhao and He, Yunzhong and Hassaan, Sami},
  institution = {Scale AI (with Phylo)},
  year = {2026},
  month = jun,
}

IEEE TSP

Sampling from Gaussian Markov Random Fields Using Stationary and Non-Stationary Subgraph Perturbations

Ying Liu, Oliver Kosut, and Alan S. Willsky

IEEE Transactions on Signal Processing, 2015

vol. 63, no. 3, pp. 576–589

DOI Bib HTML PDF

@article{liu2015sampling,
  note = {vol. 63, no. 3, pp. 576--589},
  title = {Sampling from {G}aussian {M}arkov Random Fields Using Stationary and Non-Stationary Subgraph Perturbations},
  author = {Liu, Ying and Kosut, Oliver and Willsky, Alan S.},
  journal = {IEEE Transactions on Signal Processing},
  volume = {63},
  number = {3},
  pages = {576--589},
  year = {2015},
  doi = {10.1109/TSP.2014.2375134},
}

NeurIPS

Learning Gaussian Graphical Models with Observed or Latent FVSs

Ying Liu and Alan S. Willsky

In Advances in Neural Information Processing Systems (NeurIPS), 2013

arXiv Bib HTML PDF

@inproceedings{liu2013learning,
  title = {Learning {G}aussian Graphical Models with Observed or Latent {FVS}s},
  author = {Liu, Ying and Willsky, Alan S.},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  volume = {26},
  year = {2013},
}

IEEE TSP

Feedback Message Passing for Inference in Gaussian Graphical Models

Ying Liu, Venkat Chandrasekaran, Animashree Anandkumar, and Alan S. Willsky

IEEE Transactions on Signal Processing, 2012

vol. 60, no. 8, pp. 4135–4150

arXiv Bib HTML PDF

@article{liu2012feedback,
  note = {vol. 60, no. 8, pp. 4135--4150},
  title = {Feedback Message Passing for Inference in {G}aussian Graphical Models},
  author = {Liu, Ying and Chandrasekaran, Venkat and Anandkumar, Animashree and Willsky, Alan S.},
  journal = {IEEE Transactions on Signal Processing},
  volume = {60},
  number = {8},
  pages = {4135--4150},
  year = {2012},
}

Nat. Methods

FaST Linear Mixed Models for Genome-Wide Association Studies

Christoph Lippert, Jennifer Listgarten, Ying Liu, Carl M. Kadie, Robert I. Davidson, and David Heckerman

Nature Methods, 2011

vol. 8, no. 10, pp. 833–835

DOI Bib HTML Code

@article{lippert2011fast,
  note = {vol. 8, no. 10, pp. 833--835},
  title = {{FaST} Linear Mixed Models for Genome-Wide Association Studies},
  author = {Lippert, Christoph and Listgarten, Jennifer and Liu, Ying and Kadie, Carl M. and Davidson, Robert I. and Heckerman, David},
  journal = {Nature Methods},
  volume = {8},
  number = {10},
  pages = {833--835},
  year = {2011},
  doi = {10.1038/nmeth.1681},
}