Since September 2024, I have been a Postdoctoral Research Associate in the Machine Learning Department at CMU, supervised by Pradeep Ravikumar. Previously, I was a PhD Student at MIT, where I was advised by Caroline Uhler and David Sontag.

My research focuses on theoretical and methodological foundations for trustworthy, deployable AI, informed and driven by some of the most challenging scientific and engineering applications, including drug discovery, manufacturing, and healthcare. I am particularly interested in applications that present challenges for the current AI paradigm, e.g., high-stakes applications that require safety, reliability, and/or transparency, or data-scarce applications that require extrapolation, generalization, and/or targeted data collection.

To address these challenges, I study structured/identifiable representation learning, combining tools and perspectives from causality, neurosymbolic AI, and mechanistic interpretability, and decision-theoretic learning, combining tools and perspectives from statistical learning theory, debiased/orthogonal machine learning, and (targeted) Bayesian experimental design.

(Last update: October 16th, 2026)

Service. I have served as an Area Chair at NeurIPS (2025 and 2026), Publication Chair at CLeaR (2025 and 2026), Workflow Chair at UAI (2026), and as an assistant organizer for the Identifiable Representation Learning workshop at BIRS. During my PhD, I was a member of the EECS Communication Lab, where I coached researchers on technical communication tasks such as paper writing, poster design, and oral presentation, and I organized the Causal Representation Learning Reading Gruop .

Teaching. I developed and taught a seven-lecture course on causality for MIT’s 2023 January term, check out the lecture notes and recordings.

Software. During my PhD, I developed and maintained causaldag, a Python package for creating, manipulating, and learning causal graphical models.

Some favorite quotes

• John Dewey: “Conceptions, theories and systems of thought…are tools. As in the case of all tools, their value resides not in themselves but in their capacity to work shown in the consequences of their use.”
• Unattributed: “Good mathematicians write proofs. Better mathematicians write conjectures. The best mathematicians write definitions.”
• Vladimir Vapnik: “When solving a problem of interest, do not solve a more general problem as an intermediate step.”