Hello, my name is Sarah. I am a senior at Caltech majoring in Computer Science, with minors in Mathematics and Information + Data Science, advised by Prof. Steven Low and Tony Yue Yu. I have worked with Dr. Ricardo Baptista, Profs. Adam Wierman, Steven Low and Anima Anandkumar at Caltech.
I am currently working with Dr. Benjamin Plaut (advised by Prof. Stuart Russell) at UC Berkeley's Center for Human-Compatible Artificial Intelligence.
My research interests lie in using computational mathematics to create more structured and physically grounded systems by integrating physical principles, particularly for solving complex engineering problems.
One goal in particular is developing interpretable and physically informed models and algorithms for high-dimensional, stochastic, and non-convex systems, particularly in settings where standard numerical techniques fail:
(1) How can we develop “good” models—those that are uncertainty-aware, distributionally robust, and come with provable guarantees—from data, even in data-scarce environments?
(2) How can we quantify uncertainty in tasks involving chaotic systems and complex geometric representations?
Previously, I have developed scalable probabilistic modeling algorithms for Markov Random Fields using computational measure transport.
More recently, I have also been exploring the intersection of statistical physics and machine learning, particularly by analyzing the learning process as a dynamical system.
I am also currently working on cautious learning of agentic systems in out of distribution settings.
I am always eager to discuss ideas, and welcome any comments/questions about my research interests.
