PeopleofSuperintelligence

Researcher

Peter J. Liu

Peter J. Liu is a language-model researcher whose work spans summarization, clinical note generation and Transformer training; he holds mathematics and computer science degrees from the University of Toronto, began language-model research on Google Brain in 2015, and coauthored a pointer-generator summarization paper in 2017. His 2018 work includes a clinical note-generation study and a method for generating long Wikipedia articles; he coauthored the text-to-text transfer paper that Google Research presents as T5 in 2020. In 2023, he coauthored research on Transformer training instabilities and joined the AIMO Prize Advisory Committee in 2024. He is a co-founder of Twenty Labs, whose first product is Compound, an AI financial analyst.

Now at
Twenty Labs
Works
34
Badges

Publication numbers from OpenAlex, as of 22 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2015

    Google Brain language research

    He holds degrees in mathematics and computer science from the University of Toronto. In 2015, he begins machine-learning research specializing in language models on the Google Brain team.

  2. 2017

    Pointer-generator summarization

    In 2017, he coauthors “Get To The Point: Summarization with Pointer-Generator Networks.” Its network copies source words through a pointer, and coverage tracks summarized content to discourage repetition.

  3. 2018

    Clinical note generation

    In 2018, he publishes the solo study “Learning to Write Notes in Electronic Health Records.” It predicts note content from earlier patient-record data and trains generative models on the public, de-identified MIMIC-III dataset.

  4. 2018

    Long-form Wikipedia generation

    At ICLR in 2018, he coauthors “Generating Wikipedia by Summarizing Long Sequences.” The paper frames English Wikipedia generation as multi-document summarization and introduces a decoder-only architecture for long inputs.

  5. 2019–2020

    Text-to-text transfer learning

    In 2019, he coauthors transfer-learning research that converts text-based language problems into a text-to-text format. Google Research presents the resulting model as T5 in 2020 and lists translation, summarization, question answering and classification among its tasks.

  6. 2023

    Transformer training instabilities

    In 2023, he coauthors “Small-scale proxies for large-scale Transformer training instabilities.” The study finds that some instabilities also appear in small models at high learning rates.

  7. 2024

    AIMO Prize Advisory Committee

    In 2024, he joins the AIMO Prize Advisory Committee. He is a Research Scientist at Google DeepMind, specializing in language-model research.

  8. As of October 2026

    Twenty Labs and Compound

    As of October 2026, he is a co-founder of Twenty Labs, an applied research lab whose first product is Compound, an AI financial analyst. Before co-founding the lab, he works as a Research Scientist at Google DeepMind on language modeling from pretraining through fine-tuning and post-training. His earlier production experience includes leading machine learning for Gmail Spam, deploying models to millions of users, and using neural networks for financial-fraud detection at banks.

Career

  1. NowTwenty LabsCo-founder
  2. From 2024AIMO Prize Advisory CommitteeAdvisory Committee Member
  3. From 2015Google Brain
  4. Google DeepMindResearch Scientist

Education

  1. University of Toronto

From the biography.

  1. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  2. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Google. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Peter J. Liu yet. Send us a free-licensed photo.

  1. The AIMO Prize Manager and Advisory Committee aimoprize.com. Checked 21 October 2026.
  2. Get To The Point: Summarization with Pointer-Generator Networks aclanthology.org. Checked 21 October 2026.
  3. Learning to Write Notes in Electronic Health Records arxiv.org. Checked 21 October 2026.
  4. Generating Wikipedia by Summarizing Long Sequences research.google. Checked 21 October 2026.
  5. Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer arxiv.org. Checked 21 October 2026.
  6. Exploring Transfer Learning with T5: the Text-To-Text Transfer Transformer research.google. Checked 21 October 2026.
  7. Small-scale proxies for large-scale Transformer training instabilities arxiv.org. Checked 21 October 2026.
  8. AIMO Prize Appoints Further Advisory Committee Members | AIMO Prize aimoprize.com. Checked 21 October 2026.
  9. Peter J Liu peterjliu.com. Checked 21 October 2026.
  10. About Us | Compound getcompound.ai. Checked 21 October 2026.