My interests are in scientific machine learning and computational physics. I'm particularly interested in how we represent data from physical systems, how this affects what a model can learn, and what this can reveal about the underlying physics.
My current research examines spectral representations for regional weather forecasting. My independent work explores universality, learned representations, information limits, PDE discovery, and reliable model evaluation.