UC Berkeley

The recent revolution in advanced data analytics gave rise to a growing demand among organizations for high quality data. However, in many domains such as finance and medicine, organizations have encountered obstacles in data acquisition because their target applications need sensitive data that reside across multiple parties. One promising solution to this data scarcity problem is collaborative computation, where several organizations pool together their data and compute on the joint dataset. This type of computation enables parties to acquire a larger volume of data, as well as more diverse data. Unfortunately, organizations are often unwilling or unable to share their data in plaintext due to business competition or government regulation.

My dissertation focuses on solving this problem by enabling organizations to run complex computations on the joint dataset without revealing their sensitive input to the other parties. This dissertation presents four systems that utilize hardware enclaves as well as advanced cryptographic techniques for secure computation on workloads that range from SQL analytics to machine learning. By utilizing a wide range of tools from both systems and cryptography and also innovating on them, my systems provide strong and provable security guarantees and are often orders of magnitude faster compared to prior work or the more straightforward ways of integrating cryptography into systems.