Research

TDEs provide a rare opportunity to observe black hole accretion turn on in an otherwise quiescent galaxy. My work focuses on multi-wavelength observations of individual events, connecting their photometric and spectroscopic evolution to accretion disk geometry and black hole properties.

Distinguishing TDEs from AGN in photometric surveys is genuinely difficult, and misclassification affects the science of both populations. My work connects nuclear transient behavior to host-galaxy properties, with particular attention to post-starburst galaxies, and develops classifiers for separating the two populations in survey data.

Fitting physically motivated spectral models to TDE and AGN data requires robust Bayesian inference with well-characterized uncertainties. My work in this area centers on FEADME, a JAX/NumPyro framework for fitting relativistic accretion disk profiles to double-peaked emission lines on GPU hardware.

Rubin/LSST will produce far more nuclear transient alerts than can be followed up spectroscopically, requiring classification from photometric light curves alone. My work develops neural process and transformer-based models for early-time classification on irregularly sampled, multi-band data, with a particular focus on TDE identification.

Before my PhD, I worked on the circumgalactic medium (the diffuse gas surrounding galaxies) using UV absorption spectroscopy. I developed Spectacle, a line-analysis package for absorption spectra, and used it to study how simulation resolution affects CGM kinematics, comparing simulated absorbers directly to HST/COS observations.