PySpark

Built and maintained by a team of differential privacy experts, and running in production at institutions like the U.S. Census Bureau. Runs on Spark and effortlessly supports input tables containing billions of rows. Supports a large and ever-growing list of aggregation functions, data transformation operators, and privacy definitions. Perform public and private joins, filters, or user-defined functions on your data. Compute counts, sums, quantiles, and more under multiple privacy models. Differential privacy is made easy, thanks to our simple tutorials and extensive documentation. Tumult Analytics is built on our sophisticated privacy foundation, Tumult Core, which mediates access to sensitive data and means that every program and application comes with an embedded proof of privacy. Built by composing small, easy-to-review components. Provably safe stability tracking and floating-point primitives. Uses a generic framework based on peer-reviewed research.