Johan Leduc and Nicolas Grislain, Sarus Technologies
Working with privacy-sensitive data today, whether it is in health-care, insurance or any other industry, is a complex and slow process often involving long manual reviews by compliance teams. The recent development of differential privacy helped standardize what privacy protection means. As such it has the potential to unlock the automation and scaling of data analysis on privacy sensitive data. To help realize this promise, we designed and built a framework in which an analyst can write data analysis jobs with common data-science tools and languages: SQL, numpy, pandas, scikit-learn, and have them compiled into differentially private jobs executed remotely on the sensitive data. In this talk, we will describe how a user expresses his job declaratively in python and how his python code is analyzed and compiled, before it is run and a result is eventually returned.
Johan Leduc, Sarus
Johan Leduc is a senior data scientist at Sarus Technologies. He graduated from Ecole Polytechnique in 2014. He started his career in the energy sector and switched to data science in 2019. He joined Sarus (YC W22) in 2020 as the first employee and has been working on private synthetic data generation and private data analysis.
Nicolas Grislain, Sarus
Nicolas Grislain is Chief Science Officer at Sarus Technologies. He graduated from École Normale Supérieure de Lyon in Mathematics and Computer Science. Nicolas started his career in economics and finance modeling at the French Treasury and then at Société Générale. He co-founded a first company: AlephD, in 2012, where he was also leading Research and Development. AlephD was acquired by Yahoo in 2016. In 2020 he co-founded Sarus Technologies (YC W22) with the same founding team as AlephD.
author = {Johan Leduc and Nicolas Grislain},
title = {Compiling Python Programs into Differentially Private Ones},
year = {2022},
address = {Santa Clara, CA},
publisher = {USENIX Association},
month = jun
}
