PeopleofSuperintelligence

Researcher

Jason Rute

Jason Rute is an AI researcher whose work applies machine learning to formal mathematics and theorem proving. He earned a mathematical sciences PhD at Carnegie Mellon in 2013 and took his first postdoctoral position in Penn State's Department of Mathematics. A 2018 arXiv survey linked algorithmic randomness with computable measure theory; he coauthored PACT in 2021 and Graph2Tac in 2024. At IBM Research in January 2025, he presented a talk on practical AI theorem provers, and in June 2025 he coauthored Magistral, Mistral's first reasoning model. His August 2026 Leanstral paper presented a series of generalist code-agent models for Lean 4; as of October 2026, he is an AI researcher at Mistral AI focused on teaching AI systems to reason, especially in mathematics.

Badges

Timeline

  1. 2013

    A PhD and postdoc

    He completes a PhD in mathematical sciences at Carnegie Mellon in 2013, with a thesis on algorithmic randomness and computable analysis. His first postdoctoral position is as a postdoctoral assistant professor in Penn State's Department of Mathematics.

  2. 2018

    Algorithmic randomness and measure theory

    In 2018, he posts an arXiv survey on constructive and computable measure theory, with emphasis on its connections to algorithmic randomness.

  3. 2021

    Proof artifact co-training

    In 2021, he coauthors “Proof Artifact Co-training for Theorem Proving with Language Models” (PACT), which trains a language model for Lean theorem proving with self-supervised data from kernel-level proof terms. It improves success on a held-out theorem set from 32% to 48%.

  4. 2024

    Graph2Tac for theorem proving

    At ICML in 2024, he coauthors Graph2Tac, which builds hierarchical representations of mathematical definitions for proof assistants. Its online definition task solves 1.5 times as many theorems as an offline baseline.

  5. January 2025

    The last mile at IBM

    In January 2025, while at IBM Research, he presents 'The last mile,' a talk on making AI theorem provers useful for everyday proof-assistant users.

  6. June 2025

    Mistral's first reasoning model

    In June 2025, he coauthors Magistral, introduced as Mistral's first reasoning model. The paper describes a scalable reinforcement-learning pipeline built with Mistral's own models and infrastructure.

  7. August 2026

    Leanstral for Lean 4

    In August 2026, he coauthors Leanstral, a series of generalist code-agent models for Lean 4.

  8. As of October 2026

    AI research at Mistral

    As of October 2026, he is an AI researcher at Mistral AI. He focuses on teaching AI systems to reason, especially in mathematics.

Career

  1. NowMistral AIAI researcher
  2. From 2025IBM Research
  3. From 2013Pennsylvania State University

Education

  1. 2013Carnegie Mellon UniversityPhD

From the biography.

  1. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes Jason Rute. 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.
  2. 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.
  3. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Mistral AI. 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.
  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 Jason Rute yet. Send us a free-licensed photo.

  1. Past Ph.D. Recipients - Mathematical Sciences - Mellon College of Science - Carnegie Mellon University cmu.edu. Checked 21 October 2026.
  2. Pure and Applied Logic logic.cmu.edu. Checked 21 October 2026.
  3. [1812.03375] On the close interaction between algorithmic randomness and constructive/computable measure theory arxiv.org. Checked 21 October 2026.
  4. [2102.06203] Proof Artifact Co-training for Theorem Proving with Language Models arxiv.org. Checked 21 October 2026.
  5. Graph2Tac: Online Representation Learning of Formal Math Concepts proceedings.mlr.press. Checked 21 October 2026.
  6. The last mile: How do we make AI theorem provers which work in the real world for real users and not just on benchmarks? researchseminars.org. Checked 21 October 2026.
  7. [2506.10910] Magistral arxiv.org. Checked 21 October 2026.
  8. Leanstral | alphaXiv alphaxiv.org. Checked 21 October 2026.
  9. Jason Rute | Website of Jason Rute jasonrute.github.io. Checked 21 October 2026.