Led the release of 'The Well': the largest collection of diverse physics simulation datasets (from biology to astrophysics), published at NeurIPS 2024 (Datasets & Benchmarks track).
Developed a blind denoising algorithm using Gibbs diffusion with parameter inference and uncertainty quantification, accepted at ICML 2024.
Published MoMo (Momentum Models for Adaptive Learning Rates) at ICML 2024.
Published PAC-Bayesian analysis of adaptive sliced-Wasserstein distances at ICML 2023.
Published work on photonic differential privacy at NeurIPS 2021.
Multiple papers in NeurIPS workshops (AI4Science, MLPS), including AstroCLIP, xVal, and CMB dust removal with diffusion models.
Published research on photonic computing and kernel methods in Optics Express and AISTATS 2023.