Bradley Hutchinson

My research interests center on advancing mathematical and computational methods for machine learning, with an emphasis on developing new frameworks that improve interpretability, constrain model expressivity, and enable robust scientific inference. I am particularly interested in using astrophysics as a rich application domain for advancing computational science and AI, while building methods that extract scientifically meaningful structure from large, complex observational datasets.

My current work focuses on applying unsupervised and representation learning techniques to large-scale multiband photometric surveys to identify rare stellar populations and characterize astrophysical phenomena such as interstellar reddening.

More broadly, I am interested in computational science approaches that use machine learning not as a purely black-box predictor, but as a physically informed tool for scientific discovery. My long-term goal is to develop scalable methods that can efficiently analyze next-generation datasets, uncover rare or previously inaccessible phenomena, and strengthen the connection between simulations, theoretical models, and observations. My background in astronomy, physics, mathematics, and scientific computing has prepared me to contribute to interdisciplinary computational science research.