Human Judgment in Technical Systems
Technical systems are built, trained, used, and interpreted by people. I study how human judgment and institutional context shape the behavior of data-driven systems, especially when those systems appear more neutral or objective than they really are.
Selective Inheritance of Bias
Machine learning systems can inherit and amplify human bias when they are trained on decisions made under pressure, uncertainty, or unequal social conditions. This work studies how biased human choices can become embedded in algorithmic systems and spread across institutional or online environments. (WP 1, WP 2)
Public-Interest Data Practice
Through the Public Policy Data Lab, I study and build the decision infrastructure of public institutions. The clinic trains students to build useful technical systems in real institutional contexts while critically examining what those systems see, miss, and change. (FAccT 2026)
Media, Platforms, and Manipulation
Online platforms shape behavior through ranking systems, recommendation algorithms, headlines, and interface design. This work studies the gap between what people can recognize as credible or manipulative and what actually changes their attention, choices, and behavior. (Ongoing)