PeopleofSuperintelligence

Researcher

Thibaut Lavril

Thibaut Lavril is a researcher whose work spans computational catalysis, automated theorem proving and language models. He joined Meta's FAIR group as an AI research engineer in 2019 and coauthored the Open Catalyst 2020 dataset, HyperTree Proof Search and LLaMA. He moved to Mistral in June 2023, where he coauthored Mistral 7B and Mixtral. In May 2025, he was an AI research engineer at Mistral AI; in June 2025, he coauthored Magistral, which introduced Magistral Medium trained on Mistral Medium 3 with reinforcement learning alone.

Timeline

  1. February 2019–June 2023

    AI research engineer at FAIR

    He joins Meta's FAIR group as an AI research engineer in February 2019. He leaves Meta in June 2023 after four years and five months.

  2. October 2020

    The Open Catalyst 2020 dataset

    The Open Catalyst 2020 dataset paper, which he coauthors in October 2020, describes 1,281,040 density-functional-theory relaxations across materials, surfaces and adsorbates.

  3. May 2022

    HyperTree Proof Search

    In May 2022, he coauthors HyperTree Proof Search, an online training approach for a transformer-based automated theorem prover that learns from previous proof searches.

  4. February 2023

    LLaMA's foundation models

    In February 2023, he coauthors LLaMA, a family of foundation language models ranging from 7B to 65B parameters, trained on trillions of tokens with publicly available datasets.

  5. June 2023

    Joining Mistral

    In June 2023, he leaves Meta and joins Mistral as an AI research engineer.

  6. October 2023

    Mistral 7B

    Mistral 7B, which he coauthors in October 2023, is a 7-billion-parameter language model that uses grouped-query attention for faster inference and sliding-window attention to handle long sequences.

  7. January 2024

    Mixtral's sparse experts

    Mixtral 8x7B, which he coauthors in January 2024, is a sparse mixture-of-experts model with eight feedforward blocks in each layer and a router that selects two experts for every token.

  8. May 2025–June 2025

    Mistral engineer and Magistral

    In May 2025, he is an AI research engineer at Mistral AI. In June 2025, he coauthors Magistral, a paper that presents Magistral Medium, trained on top of Mistral Medium 3 with reinforcement learning alone.

Career

  1. From 2023Mistral AIAI research engineer
  2. 2019–2023FAIR (Meta)AI research engineer

From the biography.

  1. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Thibaut Lavril. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  2. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Thibaut Lavril. 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.
  3. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Mistral AI. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  4. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Mistral AI. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.

Colleagues

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

Sources and credits

Photo: We have no published photo of Thibaut Lavril yet. Send us a free-licensed photo.

  1. 两岁的Llama，最初的14位作者已跑了11个！Mistral成最大赢家_澎湃号·湃客_澎湃新闻-The Paper thepaper.cn. Checked 21 October 2026.
  2. Top Llama AI Researchers Depart Meta to Join Rivals, Notably French Startup Mistral hyper.ai. Checked 21 October 2026.
  3. The Open Catalyst 2020 (OC20) Dataset and Community Challenges arxiv.org. Checked 21 October 2026.
  4. HyperTree Proof Search for Neural Theorem Proving arxiv.org. Checked 21 October 2026.
  5. LLaMA: Open and Efficient Foundation Language Models arxiv.org. Checked 21 October 2026.
  6. Mistral 7B arxiv.org. Checked 21 October 2026.
  7. mistralai/Mistral-7B-v0.1 · Hugging Face huggingface.co. Checked 21 October 2026.
  8. Mixtral of Experts arxiv.org. Checked 21 October 2026.
  9. Magistral arxiv.org. Checked 21 October 2026.