Kareem Hegazy
Kareem Hegazy
Postdoctoral Researcher · UC Berkeley · Lawrence Berkeley National Lab · ICSI
I am a postdoctoral researcher at UC Berkeley, Lawrence Berkeley National Lab, and ICSI, working with Michael Mahoney and Benjamin Erichson at the intersection of Physics, Chemistry, and AI. I received my Physics PhD from Stanford in 2023 under Phil Bucksbaum and Ryan Coffee, where I studied excited-state quantum molecular dynamics through ultrafast gas-phase diffraction. In summer 2019 I was an AI Research Fellow at Google X on the Blueshift team. I previously earned an Honors B.Sc. in Physics and Math from the University of Michigan, where I researched the Higgs Boson at CERN with Bing Zhou.
I am broadly interested in the intersection of physics and machine learning — LLMs for time series, ML for thermodynamic processes and PDE modeling, ML chemical potentials, and inverse problems. </div>
Selected Publications
Neural Equilibria for Long-Term Prediction of Nonlinear Conservation Laws
Under review at ICML
Physics–ML hybrid framework grounded in Boltzmann-BGK kinetic theory — stable forecasts for shock propagation and turbulence at Re=50,000.
Powerformer: A Transformer with Weighted Causal Attention for Time-Series Forecasting
AISTATS 2026
State-of-the-art transformer for time-series forecasting with weighted causal attention that encodes recency bias and local structure.
The False Promise of Zero-Shot Super-Resolution in Machine-Learned Operators
ICLR 2026
Demonstrates fundamental limitations of zero-shot super-resolution in ML operators for PDEs, challenging a widely held assumption in the field.
Neural Network Path Optimization for Finding Transition States on a Machine Learned Potential
Submitted to Nature Computational Science
A double-ended neural-network framework for chemical transition state search on ML interatomic potentials — outperforms NEB / GSM at high-throughput scale.
Tracking Dissociation Pathways of Nitrobenzene via MeV Ultrafast Electron Diffraction
J. Phys. B, 57, 195101 (2024)
Resolving photoinduced dissociation pathways of nitrobenzene using MeV ultrafast electron diffraction.
Applying Bayesian Inference and Deterministic Anisotropy to Retrieve the Molecular Structure from Gas-Phase Diffraction Experiments
Communications Physics, 6, 325 (2023)
Novel Bayesian statistical framework that transforms gas-phase diffraction into 3D molecular images — 100× resolution improvement, no simulation bottleneck.
For a full list, see my Google Scholar or the full publications page.
Selected Blogs
Blog posts coming soon.
Curriculum Vitae
Get in touch
Email: khegazy@berkeley.edu
GitHub · LinkedIn · Twitter · Google Scholar