About

I’m Sonia, a second-year PhD student in Prof. Frances Anold’s lab. My research uses machine learning to accelerate directed evolution in order to engineer enzymes that catalyze new-to-nature chemistry. Specifically, I work on guiding protein language models with experimental data to generate novel enzymes using a reinforement learning technique known as direct preference optimization (DPO).

Previously, I studied biochemistry and computer science at the University of Chicago where I worked with Prof. Rama Ranganathan to design proteins using Potts models parameterized by a protein’s epistasis terms, and was the president of UChicago’s iGem team, Genehackers. Before joining the Arnold lab, I rotated with Michael Elowitz to elucidate the combinatorial effect of cytokines on the Jak/STAT pathway and Matt Thomson where I examined the sequence statistics of proteins designed via foldtuing, an algorithm that generates proteins to be sequentially novel yet structurally degenerate to known folds.

Overall, I want to acheive precise control and engineerability of biological systems from the molecular to the organismal scale, and I am excited about the application of ML in biology to acheive that vision. Please reach out if you are interested to talk or collaborate!