"Toward Secure and Trustworthy Wireless Sensing:" Wireless sensing uses radio signals to perceive physical properties of people and their environments, enabling applications such as remote health monitoring, fall detection, and elderly care without the intrusiveness of cameras. But the apparent privacy advantage of radio no longer holds: the same signals readily leak sensitive biometric information such as breathing, heart rate, and gait, and the problem deepens when these systems fail to respect appropriate information flows. The upcoming 6G standard proposes to integrate sensing directly into communication infrastructure, turning every base station into a potential observer, but also offering a rare chance to build privacy guarantees before deployment. In this talk, I will explore how user privacy can be protected in this new regime, from user-side opt-in defenses to end-to-end secure system design. — Tanvir Ahmed

"The Structure and Interpretation of Socially Embedded Algorithmic Outputs": Applying AI algorithms to social settings requires translating model outputs into scientific findings, real-world decisions, or moral obligations. These complex social conclusions involve considerations outside of the traditional machine learning framework. In this talk, I use empirical data science, economic theory, and queer philosophy to navigate tradeoffs when estimating population-level race statistics, modeling admissions processes, and inferring missing gender data for algorithmic fairness. — Evan Dong