PeopleofSuperintelligence

Researcher

Jesse Bettencourt

Jesse Bettencourt is a PhD candidate in machine learning at the University of Toronto and the Vector Institute and a co-author of Neural Ordinary Differential Equations, which won a NeurIPS 2018 Best Paper Award. After degrees in integrated science and mathematics at McMaster University and in mathematics at the University of Toronto, he completed a computer science MSc co-supervised by David Duvenaud and Roger Grosse. With Yulia Rubanova, Duvenaud and another Toronto student, he introduced continuous-depth models in 2018 that parameterize the derivative of a hidden state with a neural network and compute the output with a black-box differential equation solver, and he later co-authored the DiffEqFlux.jl library and a 2020 paper on learning differential equations that are easy to solve. From January 2025 to January 2026 he was a research scientist intern in NVIDIA's Spatial Intelligence Lab, where he led CARV, a variance-reduction method for diffusion-teacher gradients. His research interests include neural differential equations, automatic differentiation, gradient estimation and generative modeling.

Timeline

  1. 2011–2016

    Mathematics at McMaster and Toronto

    He studies for a BSc in Integrated Science and Mathematics at McMaster University from 2011 to 2015, then for an MSc in Mathematics at the University of Toronto from 2015 to 2016.

  2. 2017–2019

    Master's in computer science

    From 2017 to 2019 he studies for an MSc in computer science at the University of Toronto, co-supervised by David Duvenaud and Roger Grosse. He instructs CSC412/2506, Probabilistic Learning and Reasoning, in 2018 and 2019.

  3. June 2018

    Neural ordinary differential equations

    Neural Ordinary Differential Equations, written with Ricky T. Q. Chen, Yulia Rubanova and David Duvenaud, appears on arXiv in June 2018. The models parameterize the derivative of the hidden state with a neural network and compute the output with a black-box differential equation solver.

  4. 2018

    A NeurIPS Best Paper Award

    Neural Ordinary Differential Equations wins one of four Best Paper awards at NeurIPS 2018. He also shares Best Student Paper at the Advances in Approximate Bayesian Inference Workshop in 2018 with Will Grathwohl and Ricky T. Q. Chen for FFJORD, a continuous-time generative model written with Ilya Sutskever and David Duvenaud.

  5. 2019

    Neural ODEs in Julia and JAX

    He is one of the authors of DiffEqFlux.jl, a Julia library for neural differential equations, described on arXiv in February 2019. At JuliaCon in July 2019 he gives a lightning talk on neural ODEs implemented in DiffEqFlux.jl, and the NeurIPS 2019 Program Transformations workshop carries his paper with Matthew J. Johnson and David Duvenaud on Taylor-mode automatic differentiation in JAX.

  6. 2020

    Learning easy-to-solve differential equations

    He co-authors Learning Differential Equations that are Easy to Solve with Jacob Kelly, Matthew James Johnson and David Duvenaud, published at NeurIPS 2020. The paper introduces a differentiable surrogate for the time cost of numerical solvers, computed with Taylor-mode automatic differentiation, so that models trade performance against the cost of solving the learned dynamics.

  7. January 2025–January 2026

    Leading CARV at NVIDIA

    From January 2025 to January 2026 he is a research scientist intern in NVIDIA's Spatial Intelligence Lab. There he leads CARV, a compute-aware variance-reduction framework for diffusion-teacher gradients that gives 2 to 3 times effective compute multipliers in 3D generation and data attribution experiments, and the paper Variance Reduction for Expectations with Diffusion Teachers is published in May 2026.

  8. As of October 2026

    PhD candidate at Toronto

    He is a PhD candidate in machine learning at the University of Toronto and the Vector Institute, supervised by David Duvenaud, and his PhD in computer science at Toronto runs from 2019. His research interests include neural differential equations, automatic differentiation, gradient estimation and generative modeling.

Career

  1. 2025–2026NVIDIAResearch scientist intern

Education

  1. 2019University of TorontoMSc in computer science
  2. From 2019University of TorontoPhD candidate
  3. From 2019Vector InstitutePhD candidate
  4. 2016University of TorontoMSc in Mathematics
  5. 2015McMaster UniversityBSc in Integrated Science and Mathematics

From the biography.

  1. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. 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.
  2. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes NVIDIA. 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.
  3. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes NVIDIA. 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.
  4. Universities (31%) and startups and other companies (25%) employ the largest shares of 2020–2026 entrantsIncludes NVIDIA. 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.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. Jesse Bettencourt cs.toronto.edu. Checked 21 October 2026.
  2. Neural Ordinary Differential Equations with DiffEqFlux :: JuliaCon 2019 :: pretalx pretalx.com. Checked 21 October 2026.
  3. [1806.07366] Neural Ordinary Differential Equations arxiv.org. Checked 21 October 2026.
  4. Awards neurips.cc. Checked 21 October 2026.
  5. Vector Researchers Win Top Honours at NeurIPS 2018 - Vector Institute for Artificial Intelligence vectorinstitute.ai. Checked 21 October 2026.
  6. [1810.01367] FFJORD: Free-form Continuous Dynamics for Scalable Reversible Generative Models arxiv.org. Checked 21 October 2026.
  7. [1902.02376] DiffEqFlux.jl - A Julia Library for Neural Differential Equations arxiv.org. Checked 21 October 2026.
  8. Taylor-Mode Automatic Differentiation for Higher-Order Derivatives in JAX | ML Anthology mlanthology.org. Checked 21 October 2026.
  9. [2007.04504] Learning Differential Equations that are Easy to Solve arxiv.org. Checked 21 October 2026.
  10. CARV: Variance Reduction for Expectations with Diffusion Teachers research.nvidia.com. Checked 21 October 2026.
  11. Variance Reduction for Expectations with Diffusion Teachers | Research research.nvidia.com. Checked 21 October 2026.