Raluca Ada Popa

Raluca Ada Popa

Robert E. and Beverly A. Brooks Assistant Professor, UC Berkeley
Department of Electrical Engineering and Computer Science
UC Berkeley
Address: 729 Soda Hall, Berkeley, CA, 94720
Email: raluca AT eecs DOT berkeley DOT edu
@ralucaadapopa




I am an assistant professor at UC Berkeley. I am interested in security, systems, and applied cryptography.
I co-founded and co-direct the RISELab, a lab aiming to build systems that are secure and intelligent.

I am also a co-founder and the CTO of PreVeil, a security startup based on my research.

Before joining UC Berkeley, I did a one-year postdoc at ETH Zürich in the System Security group led by Srdjan Capkun. Before that, in 2014, I completed my Ph.D. in computer science at MIT, my thesis being about building practical systems that compute on encrypted data. My advisor was Nickolai Zeldovich, and I was also fortunate to work closely with Hari Balakrishnan in systems, and with Shafi Goldwasser, Yael Kalai, and Vinod Vaikuntanathan in cryptography. I earned my Masters of Engineering in Computer Science in 2010 and my two Bachelors in Computer Science and Mathematics in 2009 also from MIT.

News



PhD advisees:

I have also worked closely with Sukrit Kalra and Dayeol Lee.

Postdocs:

Alumni:


My students and I focus on building secure systems with the help of modern cryptography. Instead of relying on a server that becomes a central point of attack, our research provides cryptographic guarantees of privacy or integrity even if the server is compromised. We have built a spectrum of systems within this paradigm, as follows.

Securing communication systems: Our work on WAVE and JEDI aims to bring the benefits of end-to-end encryption to challenging setups, such as publish-subscribe or access delegation systems as well as to resource constrained devices.

Securing storage systems: Popular systems for end-to-end encrypted file sharing reduce the amount of data an attacker can steal from a server. Nevertheless, the attacker can still learn significant metadata, or cause the server to misbehave (e.g., equivocate), problems that our systems Ghostor, Metal, WAVE, MiniCrypt, or Verena aim to prevent.

Securing computation systems: A large part of our work is dedicated to enabling untrusted servers to compute on sensitive data. For this, we have developed systems that can compute on encrypted data for: collaborative computation (machine learning: Helen, Delphi, Visor, Bost et al.; analytics: OCQ, Opaque), databases (Oblix, Opaque, Arx), network processing (SafeBricks, Embark, BlindBox), and general purpose programs (DIZK, Civet). Our works explore two separate approaches, secure multi-party computation and hardware enclaves (combined with cryptography), because of their tradeoffs. For our systems using hardware enclaves, we develop oblivious algorithms, which eliminate a large class of side-channel attacks (e.g., our Membuster attack).
Here is a talk on our secure collaborative computation work:

Papers

Theses



CS 161: Computer Security, Spring 2020, undergraduate, ≈620 students

CS 294: Decentralized Security: Theory and Systems, Fall 2019, graduate, NEW COURSE

CS 161: Computer Security, Spring 2019, undergraduate, ≈600 students

CS 261: Security in Computer Systems, Fall 2018, graduate

CS 161: Computer Security, Spring 2018, undergraduate, ≈625 students

CS 261: Security in Computer Systems, Fall 2017, graduate

CS 161: Computer Security, Fall 2016, undergraduate, ≈380 students

CS 294: RISE Lab: Real-time, Intelligent, and Secure Systems, Fall 2016, graduate, NEW COURSE

CS 161: Computer Security, Spring 2016, undergraduate, ≈500 students

CS 261: Security in Computer Systems, Fall 2015, graduate


As professor:

As graduate student:

As undergraduate student:





DARE

In early 2019, I launched , a diversity program aiming to match strong undergraduates, many from underrepresented backgrounds, with EECS professors for research. With help from EECS, we developed a web application system for DARE. As of June 2020, 30 professors are part of DARE, and a total of about 41 undergraduate students, many from underrepresented backgrounds, have engaged in research with EECS faculty.

Workshop

I am a Program Chair and organizer of CCS 2020 workshop PPMLP: Privacy-preserving machine learning in practice.

PC committees