I’ve recently relocated to Boston, to start a postdoc at MIT. If you’re in the Boston area and are interested in talking about issues or research projects in the general area of algorithmic fairness, data privacy and reconstruction attacks, or redistricting, please feel free to send a message to zschutzman [at] gmail [dot] com.
At MIT, I am a Michael Hammer Postdoctoral Fellow in the Institute for Data, Systems, and Society with an appointment in the Schwartzman College of Computing’s intiative for Social and Ethical Responsibilities of Computing.
I recently completed my PhD in the CIS Department at the University of Pennsylvania, advised by Aaron Roth. I like to think about game theory, learning theory, algorithmic fairness, and computational social science and how strategic agents, algorithmic processes, and social norms come together and interact, both mathematically and practically.
My work examines social considerations of algorithmic outputs in a variety of contexts, including how differentially private census statistics might offer different levels of privacy protection for members of different demographic groups, the design of consumer financial products which can fairly serve customers with diverse risk tolerances, and the effects of using automated procedures to draw electoral districts on the representativeness of the elected slate of candidates.
At Penn, I was affiliated with the CS Theory Research Group, the Warren Center for Network & Data Sciences, and the Penn Research in Machine Learning (PRiML) group. I have also worked with the Metric Geometry and Gerrymandering Group at MIT and Tufts and was on the faculty for their 2019 Voting Rights Data Institute summer program.
I have moved to Boston and will be starting as a postdoc at MIT IDSS this September!
Travis Dick, Matthew Joseph, and I ran successful attacks against Diffix, a purportedly privacy-preserving data analysis system. I contributed to a pair of blog posts on the theoretical basis for our attacks and the practical details of executing the attacks and what kinds of information we could extract.