PeopleofSuperintelligence

Lab leader

Sara Hooker

Sara Hooker is a machine-learning researcher and lab leader whose work links model efficiency with wider access to research and language technology. She founded Delta Analytics, spent five years as a Google DeepMind research scientist, and led Cohere’s AI research group from 2022 to 2025. Her co-authored Aya paper describes an instruction-following model for 101 languages and received ACL’s 2024 Best Paper Award. She also wrote about compute thresholds as a tool for AI governance, and TIME named her to its 2024 TIME100 AI list. In 2025, she co-founded Adaption Labs, where her 2026 research tested AI-generated checklists for training data.

Now at
Adaption Labs
Works
120
Citations
3,049
h-index
21

Publication numbers from OpenAlex, as of 14 October 2026.

Photo of Sara Hooker

Timeline

  1. 2010s

    Data for social good

    She founded Delta Analytics, a nonprofit whose goal was to build technical capacity around the world.

  2. 2020

    When hardware picks winners

    Her essay “The Hardware Lottery” names a pattern in which research ideas win when they suit available hardware and software. It shows how hardware constraints can delay research progress.

  3. 2022–2025

    Leading Cohere For AI

    Her background includes a computer-science PhD at Mila, supervised by Hugo Larochelle and Aaron Courville, and five years as a Google DeepMind research scientist. In 2022, she joined Cohere to lead Cohere Labs and, by 2024, held the titles VP of Research and Head of Cohere For AI. She left Cohere in September 2025.

  4. July 2024

    Questioning compute thresholds

    Her essay “On the Limitations of Compute Thresholds as a Governance Strategy” argues that current compute thresholds are shortsighted and likely to fail to mitigate risk. It describes the connection between compute and risk as uncertain and rapidly changing.

  5. August 2024

    A model across 101 languages

    She co-authored the Aya Model paper. It describes a generative language model that follows instructions in 101 languages, more than half lower-resource, and received ACL’s 2024 Best Paper Award.

  6. September 2024

    On TIME’s AI list

    TIME named her to its 2024 TIME100 AI list in the “Shapers” category.

  7. October 2025

    A new adaptive AI company

    After leaving Cohere, she co-founded Adaption Labs with former colleague Sudip Roy. The startup set out to build efficient AI systems that adapt and learn continuously.

  8. October 2026

    Checklists for training data

    She co-authored an Adaption study of an AI agent that creates checklists to filter training data. The tests found higher win rates across general, medical, legal, and technology domains for models trained on checklist-filtered examples.

Sources and credits

Photo: sarahooker.me; Sara Hooker - Home; creator not stated. License: Not stated by source. Source: sarahooker.me. Cropped to a square and resized. Ask us to remove or replace this photo.

  1. Educational Outreach - Sara Hooker sarahooker.me. Checked 14 October 2026.
  2. The Hardware Lottery arxiv.org. Checked 14 October 2026.
  3. Sara Hooker - Home sarahooker.me. Checked 14 October 2026.
  4. Ex-Cohere execs Sara Hooker and Sudip Roy unveil new AI startup betakit.com. Checked 14 October 2026.
  5. Sara Hooker ’13 named to TIME100 list of most influential people in AI carleton.edu. Checked 14 October 2026.
  6. On the Limitations of Compute Thresholds as a Governance Strategy arxiv.org. Checked 14 October 2026.
  7. Aya Model: An Instruction Finetuned Open-Access Multilingual Language Model aclanthology.org. Checked 14 October 2026.
  8. Automated Data Quality Control With Agentic Checklists adaptionlabs.ai. Checked 14 October 2026.