Percy Liang

Percy Liang is a Professor of Computer Science, with a courtesy appointment in Statistics, at Stanford University. He received his B.S. in 2004 and M.Eng. in 2005 from the Massachusetts Institute of Technology, and his Ph.D. in 2011 from the University of California, Berkeley, advised by Michael Jordan and Dan Klein. After completing a post-doctoral fellowship at Google in 2012, he joined the Stanford Computer Science faculty, where he progressed from assistant to associate and then full professor. Liang serves as the Director of the Stanford Center for Research on Foundation Models (CRFM), focusing on advancing the understanding and accessibility of foundation models through open-source initiatives and rigorous benchmarking. He is also known for developing CodaLab Worksheets, a platform for reproducible research that tracks experiment provenance from raw data to results. Additionally, he leads Marin, an effort to build AI models with radical openness, including preregistered experiments and live engagement.

His research specializes in machine learning and natural language processing, encompassing robustness, interpretability, human interaction, learning theory, grounding, semantics, reasoning, and particularly foundation models such as language models. Key publications include 'Prefix-Tuning: Optimizing Continuous Prompts for Generation' (Li and Liang, ACL 2021), 'WILDS: A Benchmark of in-the-Wild Distribution Shifts' (Koh et al., JMLR 2021), 'Stronger Data Poisoning Attacks Break Data Sanitization Defenses' (Koh, Steinhardt, and Liang, Machine Learning 2021), 'Concept Bottleneck Models' (Koh et al., ICML 2020), and 'Robust Encodings: A Framework for Combating Adversarial Typos' (Jones et al., ACL 2020). Liang has received major awards including the Presidential Early Career Award for Scientists and Engineers (2019), IJCAI Computers and Thought Award (2016), NSF CAREER Award (2016), Sloan Research Fellowship (2015), and Microsoft Research Faculty Fellowship (2014), along with best paper awards at ACL, EMNLP, ICML, COLT, ISMIR, CHI, UIST, and RSS. He has mentored over 30 Ph.D. students and postdocs, many now faculty at institutions like UC Berkeley, Stanford, NYU, and researchers at OpenAI, Google DeepMind, and Meta. Liang teaches courses including Language Models from Scratch (CS336), Advances in Foundation Models (CS324), and Artificial Intelligence: Principles and Techniques (CS221).

Professional Email: pliang@cs.stanford.edu