Percy Liang
Percy Liang is a professor of computer science at Stanford UniversityStanford UniversityStanford University is a research university in California founded by Leland and Jane Stanford in 1885 and opened in 1891.Open the full entry, director of its Center for Research on Foundation Models and a co-founder of Together AI. He studies natural language processing and machine learning. His projects include HELM, a framework for evaluating language models, and Marin, which opens model experiments to public scrutiny from registration through results. He also trained as a classical pianist.
Contents7 sections
Key facts
Early life: High school in Arizona and classical piano
Liang began piano lessons at eight and attended high school in ArizonaArizonaArizona is a state in the southwestern United States; its capital and largest city is Phoenix. Arizona was admitted to the Union on February 14, 1912, as the 48th state. Its population at the 2020 Census (April 1, 2020) was 7,151,502.Open the full entry. He won the top prize in the Phoenix Young Musicians Competition in both 1999 and 2000. At MITMassachusetts Institute of TechnologyThe Massachusetts Institute of Technology (MIT) is a private research university in Cambridge, Massachusetts, United States. It received its charter from Massachusetts in 1861, admitted its first students in 1865 and moved from Boston to its present campus in 1916. As of October 2025 it had 11,816 students, and it counts 106 Nobel laureates among members of its community.Open the full entry he majored in computer science, minored in music, joined chamber groups, tried singing and played a sailor in Gilbert and Sullivan's Ruddigore. His March 2004 recital at Killian Hall featured a Mozart sonata in C major, two pieces by MIT composer Peter Child and Liszt's B-minor sonata. He won MIT's 2004 Concerto Competition and the adult division of KDFC's classical music competition in 2008.[1][2]
2004–2019Language research and reproducible experiments
Liang earned his MIT bachelor's degree in 2004 and completed an MEng under Michael CollinsMichael Collins (astronaut)Michael Collins (1930–2021) was an American astronaut and Air Force officer who served as command module pilot of the Apollo 11 mission in 1969. After leaving NASA he was U.S. Assistant Secretary of State for Public Affairs and director of the National Air and Space Museum.Open the full entry the following year. He received his doctorate at UC BerkeleyUniversity of California, BerkeleyThe University of California, Berkeley (UC Berkeley) is a campus of the University of California system; the University of California was founded in 1868 through a merger of two institutions and moved to Berkeley, California, in 1873.Open the full entry in 2011 under Michael JordanMichael I. JordanMichael I. Jordan is an American computer scientist and statistician, a UC Berkeley professor emeritus and research director at Inria in France, known for work on statistical machine learning, variational inference and probabilistic graphical models; he won the inaugural WLA Prize and a BBVA Frontiers of Knowledge Award.Open the full entry and Dan Klein, then held a postdoctoral position at GoogleGoogleGoogle is an American technology company founded by Larry Page and Sergey Brin and incorporated in 1998. It became an Alphabet subsidiary in 2015. Its businesses include internet search, advertising, software platforms, cloud computing, and artificial intelligence.Open the full entry in 2012. He studied how computers learn the structure and meaning of language from data and created CodaLab Worksheets to trace experiments from raw data through final results. He received a Sloan Research Fellowship in 2015 and a US Presidential Early Career Award for Scientists and Engineers in 2019.[2][3]
From 2022Evaluating foundation models, transparency and teaching
At Stanford, Liang holds the Thomas M. Siebel professorship and directs the Center for Research on Foundation Models. HELM, introduced in 2022, compared models across use cases on measures including accuracy, robustness and fairness, publishing the prompts and responses. In 2023 he and colleagues released the Foundation Model Transparency Index to assess what developers disclosed about training, deployment and use. He also co-founded Together AI, later launched Marin to share both successful and failed experiments, and taught Stanford's CS336 course on building language models from scratch.[3][4][5][6][2][7]
April 25, 2023Interview with No Priors
Interviewed in his Stanford office on No PriorsNo PriorsNo Priors is an AI interview podcast and media brand founded and hosted by investors Sarah Guo and Elad Gil, with episodes beginning in February 2023. It covers research, startups, infrastructure and business applications through guest interviews, host discussions and retrospectives distributed on audio platforms and YouTube.Open the full entry, Liang traced the path from computational semantics to foundation models. He explained why a single leaderboard position says little about which task a model suits, and how HELM lets readers compare multiple aspects of performance. He also discussed the questions academic and commercial labs could each pursue.[8]
He described moving HELM to a fortnightly update cycle to accommodate newly released models and the changing tasks people asked them to perform.[8]
Advice Shared on No Priors
He encouraged researchers to remain open to revising their assumptions as models change what people understand intelligence to be. (53:16)[8]
He argued that evaluating a language model should include its robustness and calibration, including whether it can signal uncertainty. (43:17)[8]
Related Organizations, Websites and Public Accounts
Stanford University: Professor of computer science and professor of statistics by courtesy. Official website: https://www.stanford.edu/ (opens in a new window).[3]
Together AI: Co-founder. Official website: https://www.together.ai/ (opens in a new window).[7]
Personal website: https://cs.stanford.edu/~pliang/ (opens in a new window).[2]
Sources8 sources
- Liang to present Advanced Music Performance recital (opens in a new window)news.mit.edu↑
- Percy Liang (opens in a new window)cs.stanford.edu↑
- Percy Liang (opens in a new window)engineering.stanford.edu↑
- Percy Liang (opens in a new window)hai.stanford.edu↑
- Holistic Evaluation of Language Models (opens in a new window)arxiv.org↑
- The Foundation Model Transparency Index (opens in a new window)arxiv.org↑
- GTC Speakers (opens in a new window)nvidia.com↑
- No Priors Ep. 7 | With Stanford Professor Dr. Percy Liang (opens in a new window)youtube.com↑