I am a physics PhD student working in the
at the University of Washington with
on machine learning in astrophysics and cosmology.
I am a member of the Dark Energy Science Collaboration
and the Information and Statistics Science Collaboration
of the Vera C. Rubin Observatory.
My current focus is on forward modeling photometric redshift (photo-z)
systematics and their impact on weak lensing cosmology,
as well as super-resolution, hyper-spectral deconvolution of Rubin images.
I use a variety of machine learning tools, including normalizing flows, variational autoencoders, and
I received a bachelor's in physics from Duke University, where I graduated summa cum
I was a
Duke Faculty Scholar
Prof. Kate Scholberg
Duke Neutrino and Cosmology Group.
Using detector simulations and Bayesian analysis, I developed data unfolding methods for the
supernova neutrino detector.
I also spent a summer at the Karlsruhe Institute of Technology, working with
Dr. Andreas Haungs
to characterize the muon content of cosmic ray air showers detected in the
In my free time I enjoy backpacking, skiing, rock climbing, and board games.
Developing a metric for evaluating observing strategies via the mutual information the observed
photometry contains about redshift.
A lower bound on the mutual information is estimated with a deep ensemble of normalizing flows
(built with my normalizing flow package, pzflow).
A python package for building normalizing flows to model the joint probability distribution of
Useful for forward modeling, data augmentation, and posterior calculation.
Using ensembles of broadband photometry to deconvolve high-resolution, photometrically calibrated
These deconvolved spectra achieve state of the art performance for template-based photo-z