Fully homomorphic encryption (FHE) is a technology that allows arbitrary computation over encrypted data without ever decrypting it. For example, users can upload encrypted data to a cloud, have the cloud process that data without the decryption key, and retrieve meaningful results without the server ever learning what the content of the encrypted data was. With privacy as a major focus across tech, FHE provides the means for users to control their data while retaining the functionality they enjoy in their apps, all in a way that is resistant against data mining, malicious actors, and data breaches.

In his talk, Ryan will cover the basics of how FHE works, the challenges in implementing a secure application in modern frameworks, and how FHE can bridge this gap. He will discuss how engineers can use the Sunscreen FHE compiler to compile Rust programs into secure programs without being cryptography experts. Finally, he will demonstrate how to implement a secure machine learning application that enables users to get valuable results without the server learning what the data or result was.