PeopleofSuperintelligence

Researcher

Felix Dangel

Felix Dangel is a machine-learning researcher whose work gives neural networks access to information beyond gradients. He studied physics at the University of Stuttgart, completed a computer-science PhD at the University of Tübingen, and co-developed BackPACK, a PyTorch library for higher-order information during training. He later coauthored Cockpit and ViViT and researched second-order optimization during a postdoctoral appointment at the Vector Institute. Since 2026, he has been an assistant professor at Concordia, an Associate Academic Member of Mila and a Canada CIFAR AI Chair.

Timeline

  1. 2012–2017

    Physics at Stuttgart

    In 2015, he earns a bachelor's degree in physics at the University of Stuttgart; in 2017, he completes a master's degree there.

  2. 2018–2023

    A PhD in computer science

    From 2018 to 2023, he completes a PhD in computer science at the Max Planck Institute for Intelligent Systems and the University of Tübingen under advisor Philipp Hennig. His thesis is Backpropagation Beyond the Gradient.

  3. 2020

    BackPACK for PyTorch

    In 2020, he coauthors BackPACK: Packing more into backprop with Frederik Kunstner and Philipp Hennig. The PyTorch library extends backpropagation to expose first- and second-order information during training.

  4. 2021–2022

    Cockpit and ViViT

    In 2021, he coauthors Cockpit, a PyTorch debugging tool that helps practitioners inspect training behavior and identify failure modes. In 2022, he coauthors ViViT, which uses the low-rank structure of the generalized Gauss-Newton matrix to compute curvature eigenvalues and per-sample directional derivatives.

  5. 2023

    A summa cum laude thesis

    In 2023, his PhD thesis earns summa cum laude, the highest doctoral distinction in the German university system.

  6. 2023–2026

    Postdoctoral researcher at Vector

    From 2023 to 2026, he is a postdoctoral researcher at the Vector Institute in Toronto, working with Yaoliang Yu and Roger Grosse. His research focuses on automatic differentiation and second-order optimization for deep learning.

  7. 2025

    Taylor-mode derivatives for PDEs

    In 2025, he coauthors the NeurIPS paper Collapsing Taylor Mode Automatic Differentiation. The method rewrites computation graphs to reduce the cost of Taylor-mode derivatives for linear PDE operators used in scientific machine learning.

  8. Since 2026

    Assistant professor and AI chair

    Since 2026, he is an assistant professor in Concordia's Department of Computer Science and Software Engineering and an Associate Academic Member of Mila. In October 2026, he becomes a Canada CIFAR AI Chair and studies second-order properties such as loss-landscape curvature.

Career

  1. Since 2026Concordia UniversityAssistant Professor
  2. Since 2026MilaAssociate Academic Member
  3. 2023–2026Vector InstitutePostdoctoral researcher

Education

  1. 2023Max Planck Institute for Intelligent SystemsPhD
  2. 2023University of TübingenPhD
  3. 2017University of StuttgartMaster's degree in physics
  4. 2015University of StuttgartBachelor's degree in physics

From the biography.

  1. Universities (31%) and startups and other companies (25%) employ the largest shares of 2020–2026 entrantsIncludes Mila. Among 819 people whose first recorded role started in 2020–2026 and who have a known current organization, 60% work in commercial sectors (25% at startups and other companies, 23% at frontier labs and 13% at big tech, with 9 in two), and 31% work at universities, compared with 66% for earlier entrants.
  2. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Mila. 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.
  3. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Mila. 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.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Mila. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.

Colleagues

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

Sources and credits

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

  1. Curriculum vitæ: Felix Dangel fdangel.com. Checked 21 October 2026.
  2. BackPACK: Packing more into backprop arxiv.org. Checked 21 October 2026.
  3. Cockpit: A Practical Debugging Tool for the Training of Deep Neural Networks arxiv.org. Checked 21 October 2026.
  4. ViViT: Curvature access through the generalized Gauss-Newton's low-rank structure arxiv.org. Checked 21 October 2026.
  5. Collapsing Taylor Mode Automatic Differentiation arxiv.org. Checked 21 October 2026.
  6. Five Mila Members Appointed Canada-CIFAR AI Chairs mila.quebec. Checked 21 October 2026.
  7. Felix Dangel | Mila mila.quebec. Checked 21 October 2026.