Tumult Platform documentation#
The Tumult Platform is a set of Python libraries for building, optimizing, and deploying programs that enforce differential privacy on their input data.
It is…
… easy to use: its interface will seem familiar to anyone with prior experience with tools like SQL or PySpark.
… feature-rich: it supports a large and ever-growing list of aggregation functions, data transformation operators, synthetic data generation methods, and privacy definitions.
… robust: it is built and maintained by a team of differential privacy experts, and runs in production at institutions like the U.S. Census Bureau.
… scalable: it runs on Spark, so it can scale to very large datasets.
The Tumult Platform has three distinct components:
Tumult Analytics: Compute statistical queries with differential privacy.
Tumult Synthetics: Generate synthetic data with differential privacy.
Tumult Tune: Optimize the privacy-utility trade-off of your programs.
This documentation introduces all of the concepts necessary to get started writing differentially private programs, without prior expertise on differential privacy. Users who wish to learn more about the fundamentals of differential privacy can consult this blog post series or this longer introduction.
Additional resources#
Contact us#
The best place to ask questions, file feature requests, or give feedback about the Tumult Platform is our Slack server. We also use it for announcements of new releases and feature additions.
Cite us#
If you use the Tumult Platform for a scientific publication, we would appreciate citations to the published software and/or its whitepaper. Both citations can be found below; for the software citation, please replace the version with the version you are using.
@software{tumultplatformdocs,
author = {Tumult Labs},
title = {Tumult Platform documentation},
month = dec,
year = 2024,
version = {latest},
url = {https://tmlt.dev}
}
@article{tumultanalyticswhitepaper,
title={Tumult {{Analytics}}: a robust, easy-to-use, scalable, and expressive framework for differential privacy},
author={Berghel, Skye and Bohannon, Philip and Desfontaines, Damien and Estes, Charles and Haney, Sam and Hartman, Luke and Hay, Michael and Machanavajjhala, Ashwin and Magerlein, Tom and Miklau, Gerome and Pai, Amritha and Sexton, William and Shrestha, Ruchit},
journal={arXiv preprint arXiv:2212.04133},
month = dec,
year={2022}
}
Privacy policy#
See our privacy policy for how your information is used.