Research:
My research revolves around the interplay between deep learning, signal processing and dynamical systems. I am broadly interested in novel deep learning methods and their application to science and engineering. I am an early contributor to the theory and practice of neural differential equations, neural operators and deep state space models. I cofounded and colead DiffeqML, a research group spanning multiple institutions (Stanford, MILA, KAIST) dedicated to deep learning and dynamical systems. Our efforts include Torchdyn, the most advanced PyTorch library for neural differential equations, and DL4DYN, a suite of datasets designed to benchmark neural operators.