My research group is at the nexus of machine learning and large astronomical surveys. Our inquiries touch on numerous areas, encompassing stellar astrophysics, star formation, galactic evolution, black holes, reionization, and cosmology. A central facet of our methodology involves expanding the horizon of Bayesian statistics with contemporary deep learning methodology. This approach proves invaluable for analyzing vast survey datasets, drawing from sources like spectroscopy (SDSS-V, DESI, 4MOST), astrometry (Gaia), photometry (Euclid, Roman, CSST), and time-series data (LSST, TESS, PLATO). This synthesis often sheds light on some of the discipline's core questions.

In the realm of machine learning, beyond simulation-based inference with both flow-based and score-based models, we're also delving into the potential of Large Language Models. I jointly head the UniverseTBD collaboration, an initiative aimed at combing through the extensive corpus of astronomical literature. By fine-tuning existing foundational models, we aspire to uncover the mechanisms behind scientific breakthroughs. On another front, we're probing the statistical behaviors of neural networks to glean mathematical revelations pertinent to our astronomical research.