Khoury College Junior Researcher Award, given to one pre-tenure faculty member each year
Research
Developing models that do more with less data.
Equivariant models for science and engineering
Networks that encode physical symmetry in their layers, built with domain scientists for fluid modeling, trajectory prediction, radar, materials and robot manipulation.
Discovering structure rather than assuming it
Relaxed equivariant networks that adapt to varying symmetry in data, and symmetry-discovery methods that find hidden invariances.
Symmetry inside the network
The symmetries of neural network parameter spaces and their effect on optimization and generalization.
Publications
An asterisk in the original listings denotes equal contribution; mathematics papers list authors alphabetically.
Honors and awards
2026
2026
Honorable Mention, IEEE Robotics and Automation Letters — five of 1,700 papers
2024
Outstanding Paper Finalist, Conference on Robot Learning
2024
Best Paper, 2nd Workshop on High-dimensional Learning Dynamics at ICLR
2015
NSF Mathematical Sciences Postdoctoral Research Fellowship
2012
Wayne C. Booth Prize for Excellence in Teaching, University of Chicago
2010
NSF Graduate Research Fellowship
Invited talks
2026
Mathematical Approaches to Challenges in Modern Artificial Intelligence, National University of Singapore
Symmetries in Neural Networks — Wallenberg Advanced Scientific Forum, Rånäs Castle, Sweden
SIAM Annual Meeting, special session on algebro-geometric approaches to deep learning, Cleveland
GraphEX Symposium, Dedham, MA
2025
Pushing the Limits of Equivariant Neural Networks — TAG-DS Workshop, San Diego
How to Design Neural Networks to Understand Algebraic and Geometric Structure — ICERM, Providence
Equivariant Neural Networks for Dynamics and Control — Woods Hole Oceanographic Institution
Leveraging Symmetry for Learning in Physical Systems — AI4Science Seminar, Georgia Tech
Pushing the Limits of Equivariant Neural Networks — EquiVision Workshop at CVPR, Nashville
Pushing the Limits of Equivariant Neural Networks — GRASP Seminar, University of Pennsylvania
Improving Convergence and Generalization Using Parameter Symmetries — AMS Special Session, JMM, Seattle
2024
Simulating Radar using Equivariant Graph Neural Networks — Recent Advances in AI for National Security, Bedford, MA
Pushing the Limits of Equivariant Neural Networks — NeurReps Global Speaker Series at MIT; Robotics and AI Institute; IROS Workshop on Equivariant Robotics, Abu Dhabi
Pushing the Limits of Equivariant Neural Networks — Caltech; AstroAI Seminar, Harvard–Smithsonian CfA; Symposium on Geometry Processing Graduate School, MIT
Improving Convergence and Generalization Using Parameter Symmetries — ICLR Oral, Vienna
Symmetry in Deep Neural Networks — Open Neighborhood Seminar, Harvard Mathematics
2023
The Strengths and Limitations of Equivariant Neural Networks — AI/ML in Physics, Georgia Tech; Widely Applied Mathematics Seminar, Harvard SEAS
Equivariant Neural Networks for Dynamics and Control — Mechanical and Aerospace Engineering Colloquium, Rutgers
IAIFI Summer School tutorial and workshop, Northeastern; Graph Exploitation Symposium, Dedham, MA
Symmetry in Deep Neural Networks — Michigan State Mathematics Colloquium; Machine Learning Seminar, UMass Amherst
Earlier talks (2015–2022) include the MIT–IBM Watson AI Lab, Los Alamos National Laboratory, the Flatiron Institute, Princeton Plasma Physics Laboratory, MIT Lincoln Laboratory, the University of Amsterdam, UC San Diego, Carnegie Mellon, Duke, Rensselaer, Boston University, and AMS sectional meetings. The full list is in the CV.