About me

I’m a third-year PhD student at UC San Diego advised by Berk Ustun. I research how we can build and understand human-centered AI using tools like uncertainty quantification, interpretability, and generative modeling. Topics I’ve been investigating lately include:

• minimizing data collection: how can we design interpretable machine learning systems that minimize data collection and maximize fairness and predictive performance? (e.g., in healthcare James et al 2022 NeurIPS TSRML Workshop)
• building data-efficient systems: how can we use domain-specific data to build more efficient and less data-hungry diffusion models? (e.g., for climate forecasting Cachay et al Preprint 2023)
• uncertainty-aware interventions: how can we leverage uncertainty quantification in scaling interpretable models? (preprint coming soon!)
• training for task relevance: how can we bias generative and graph-based models to prioritize the features human reviewers think are most important? (e.g., in forgery detection James et al ICPRAI 2022)
• fairness in NLP: how can we use distribution information and probabilistic modeling to reduce bias in NLP tasks? (e.g., James et al Neurips HCML Workshop 2020)

I earned a bachelor’s degree in Computer Science at Harvard. After graduation I worked as a Machine Learning Research Engineer, first at Lendbuzz, a startup dedicated to increasing access to financial services, and then at Twitter Cortex. Outside of research, you can find me training for my next triathlon, swimming at La Jolla Shores, and in the kitchen trying out new vegan recipes.