In today’ s world of extensive data-collection, privacy of users is ever more important. Even the so called anonymized datasets can be de-anonymized given external records. The notion of differential privacy makes concrete mathematical guarantees and is far more desirable. Through this SuperUROP project, I will be working to develop a fully scalable, fault-tolerant, automatically managed privacy-preserving machine learning framework. Such framework would be indispensable for analysts and developers, as it would allow a low-barrier, easy-to-use-and-manage solution for ensuring user privacy while providing high utility at the same time.

Through this SuperUROP project, I want to gain more experience in machine learning and privacy research, in addition to making a meaningful contribution to my research group. I’ ve taken classes in machine learning, software systems, and algorithms, and I want to expand on that knowledge with an interesting application. I also hope to publish a paper by the end of this SuperUROP experience, aiming for significant results.