My research studies data-efficient decision-making, with a particular focus on using low-cost AI simulations to reduce reliance on expensive ground-truth data. I develop statistical and optimization methods that account for the error structure of imperfect AI outputs, enabling their principled use as auxiliary information. My work spans two stages of the decision pipeline:
My research studies data-efficient decision-making, with a particular focus on using low-cost AI simulations to reduce reliance on expensive ground-truth data. I develop statistical and optimization methods that account for the error structure of imperfect AI outputs, enabling their principled use as auxiliary information. My work spans two stages of the decision pipeline:
(1) experimental design and data collection, and
(1) experimental design and data collection, and
(2) statistical inference and downstream decision-making.
(2) statistical inference and downstream decision-making.