Kaili Cao

Kaili Cao is a computational astrophysicist focused on solving the data quality crisis in next-generation space missions like the Roman Space Telescope and the James Webb Space Telescope (JWST). His core research builds the “computational bridge” between astrophysical theory and cutting-edge artificial intelligence to unlock new scientific frontiers. He specializes in physics-informed image processing to fundamentally upgrade the fidelity of observational data. During his Ph.D., he developed the Fast IMCOM algorithm, a novel formalism that processes raw, flawed data into “AI-ready” super-images with mathematically predictable properties. This work, which achieved an order-of-magnitude reduction in computational cost, is his unique technical spike. As a MIDAS Fellow, his research will leverage this foundation to deploy advanced probabilistic AI frameworks, such as BLISS, in close partnership with his mentors. With AI Mentor Dr. Camille Avestruz, he will develop a systematics-robust field-level inference pipeline for Roman, ensuring that next-generation AI is trained on rigorous, high-fidelity data to maximize the scientific return of precision cosmology programs. Concurrently, with Science Mentor Dr. Feige Wang, he will use AI to cleanly separate quasar light from host galaxies in early JWST data, solving a key mystery about the origins of supermassive black holes. His ultimate goal is to establish a robust, end-to-end data framework for the next decade of data-intensive astrophysics.