Tumult Analytics: a robust, easy-to-use, scalable, and expressive framework for differential privacy

Tumult Analytics is an open-source framework for releasing aggregate information from sensitive datasets with strong privacy guarantees, based on the proven science of differential privacy. It supports a wide range of data transformations, aggregations, and privacy models. It is deployed in production at institutions such as the IRS, the Wikimedia Foundation, and the U.S. Census Bureau.

The following whitepaper presents the main design goals of the framework, and the architecture that allows it to achieve these goals.

‍