Differential privacy is a formal model of privacy protection that has received sustained attention from the research community, whose work has shown that it is possible to reveal accurate information about a population while rigorously protecting the privacy of its constituents. While DP offers a compelling promise, organizations that choose to adopt it as their privacy standard face a number of challenges doing so.
In response to those challenges, we developed a platform that empowers an organization to perform differentially private analytics at scale. The platform is currently in use at a number of organizations, including the US Census Bureau, US Internal Revenue Service, and Wikimedia. In this talk, we will present an overview of the platform and its architecture, and also briefly describe some use cases. Components of the platform are available open-source.
We will then focus on several challenges that we faced in the design and implementation of this platform. Some of these challenges expose some surprising gaps between the theory of DP research and the practice of DP deployment and offer interesting directions for future research.
Michael Hay is the Founder/CTO of Tumult Labs and an Associate Professor of Computer Science at Colgate University. He was previously a Research Data Scientist at the US Census Bureau and a Computing Innovation Fellow at Cornell University. He holds a Ph.D. from the University of Massachusetts Amherst and a bachelor’s degree from Dartmouth College.