I am on the 2025/2026 job market!
Hello! I am a postdoc at Cornell, working with Emma Pierson and Jenna Wiens. I completed my Ph.D. in the Clinical and Applied Machine Learning group at MIT, where I was lucky to be advised by John Guttag.
I design methods to make machine learning systems robust to the imperfections of real-world data and models, with much of my work grounded in healthcare. Some specific areas I work on:
Learning from imperfect data
I develop machine learning techniques to model the human and systemic factors that shape datasets, including underreporting (NWH 2024), differential healthcare access (CHIL 2025), and financial incentives. These behavioral models identify subtle ways disparities propagate through ML systems and provide methods to mitigate these effects during model training.
I develop machine learning techniques to model the human and systemic factors that shape datasets, including underreporting (NWH 2024), differential healthcare access (CHIL 2025), and financial incentives. These behavioral models identify subtle ways disparities propagate through ML systems and provide methods to mitigate these effects during model training.
Learning from imperfect models
Many downstream applications begin with pretrained models. How can we adapt and evaluate them responsibly? My work introduces test-time interventions, such as test-time augmentation (ICCV 2021, CVPR 2025), and new evaluation methods that move beyond curated benchmarks (CHI 2023, NeurIPS 2025).
Many downstream applications begin with pretrained models. How can we adapt and evaluate them responsibly? My work introduces test-time interventions, such as test-time augmentation (ICCV 2021, CVPR 2025), and new evaluation methods that move beyond curated benchmarks (CHI 2023, NeurIPS 2025).
Bridging ML and high-stakes problems in healthcare
I connect core ML ideas to real-world challenges in women’s health, federal health insurance, and health inequality. These settings motivate new methods; for example, a new method for prevalence estimation for intimate partner violence screening (NWH 2024).
I connect core ML ideas to real-world challenges in women’s health, federal health insurance, and health inequality. These settings motivate new methods; for example, a new method for prevalence estimation for intimate partner violence screening (NWH 2024).