Somayeh Khakpash

I am a LSSTC Catalyst Fellow at Rutgers University. I work on applying AI methods to cautsic maps of quasar lensing events to extract useful represntations of these maps and connect them to the underlying lensing parameters. My research is designed to help model microlesning variability of lensed quasars in the era of the Rubin Observatory's Legacy Survey of Space and Time (LSST).
I received my PhD in Physics from Lehigh Univeristy under supervision of Professor Joshua Pepper in August 2020. After that, I was a postdoctoral researcher at university of Delaware working with Professor Federica Bianco for two years. I develop algorithms to extract features from the microlensing light curves and use the features to detect, classify and characterize the events in a fast and efficient way. As part of my postdoctoral research, I focused on analyzing light curves of supernovae, creating data-driven templates for subclasses of stripped-envelope supernovae to help classify stellar explosions photometrically and identify unusual photometric behavior.
I am the coordinator of the Rubin LSST TVS microlensing subgroup whose key activities include providing critical scientific input to Rubin Observatory to determine a final survey strategy for LSST that would benefit the microlensing science and ensure the scientific community develops tools to study microlensing in the LSST era. I was also a LSSTC Data Science Fellow and have been trained in data engineering, Bayesian statistics, machine learning, image processing, and advance data visualisation.
Furthermore, I am a member of the KELT science team, and I have taken part in sorting and vetting of exoplanet candidates. I have experience working with KELT lightcurves of M-dwarf eclipsing binaries and giant eclipsing binaries, and have used period finding algorithms to analyze these lightcurves.

The full modeling of each binary-lens microlensing event often requires significant investment of human and computing resources. And in the era of large surveys, a fast technique is required to analyze large datasets and determine planetary system candidates for follow-up observations. For that purpose, I develope algorithms to find features in simulated Roman microlensing light curves. This includes features like smoothness of the peak, symmetry, number of extrema, number of peaks, and width and height of the small deviations from the main peak. This will allow us to quickly analyze a set of microlensing light curves and later use the resulting parameters as input to machine learning algorithms to classify the events. My works mainly focuses on Roman simulated light curves, but I am also interested in the synergy between Roman Space Telescope and Rubin Observatory. I am a graduated DSFP fellow and have been trained to work on large datasets and analyze them using statistics, machine learning, data visualization, software engineering, time-domain analysis, and image processing. I am also an active member of the LSST Transients/Variables Science Collaboration working on proposing observing strategies for the Rubin Observatory. Additionally, I am a member of the KELT science team and have taken part in sorting and vetting of transiting exoplanets. Here is the list of my publications: