What can you tell us about the Sky Computing program, which is a successor to the RISELab at UC Berkeley? What is your role with the group?

I co-founded SkyLab with a few other professors at UC Berkeley after the predecessor lab RISELab (which I also co-founded), came to an end. Each Berkeley systems lab operates for five to six years and then is replaced by a successor. This happens because, after those few years, our team wants to regroup, redefine, and focus on the next major problems facing the computing industry – instead of continuing to work on older issues. The RISELab was focused on cloud systems, particularly enabling intelligent and secure systems in the cloud. The SkyLab moves the focus from the cloud to the sky of clouds to support the new generation of cross-cloud systems. In addition to being a co-founder of SkyLab, I continue to direct the security research team within this lab.

You are involved with two encrypted data or security startups, Opaque Systems and PreVeil. Can you share any tips for turning an academic software project into a commercial success?

My first tip is to work on research that addresses a severe pain point that customers or users face today. It is not enough for the technology to be mathematically sound or impressive in its capabilities; it has to address high-priority problems facing potential customers.

My second tip is to develop a research solution for the problem that is efficient while also packaging it in a very easy-to-use way. Overall, one should strive for a solution that addresses the priority pain points of clients but does not introduce any new friction for teams on the ground. In the end, the lack of added friction makes this jump from research to commercial product easier.

Confidential computing – or processing or analyzing data while it’s still encrypted – seems poised to have a big year in 2023. Why is this happening now?

Two big factors have greatly catalyzed the adoption of Confidential Computing, increased need as well as a new disruptive and available technology. On the need front, privacy regulations have only become stricter (GDPR, CCPA, and cookies being removed in advertising). On the technology front, the arrival of performant hardware enclaves from multiple hardware vendors (like Intel SGX and AMD SEV) enables high performance and confidentiality simultaneously. In recent years, the major cloud providers (Azure, Google, and AWS) have offered Confidential Computing via enclaves as a service; this removes the need for users to procure specialized hardware and make Confidential Computing available as a cloud service like any other.

Have all of the technical hurdles to confidential computing been solved? If not, where are the biggest hurdles today?

Like any other technology, Confidential Computing is not perfect. People are still working on eliminating side channels even with this technology around. My research group made significant progress through a technique called “in-enclave oblivious computation.” This helped to progress the solution because a program that runs with oblivious computation inside an enclave protects itself from a wide range of side-channel attacks. When running enclaves on a major cloud with state-of-the-art “traditional” security, the result is a strong security solution with two different layers of security, which we researchers call “defense-in-depth.”

Another big hurdle, albeit less technical, is the amount of education still needed throughout the industry. While Confidential Computing enables privacy-preserving computation with a wide range of applications, many users do not yet understand the powerful capabilities it has.

Outside of the professional sphere, what can you share about yourself that your colleagues might be surprised to learn – any unique hobbies or stories?

I come from a medieval town in Transylvania, close to Dracula’s castle, which is interesting because: I focus best on work after midnight, I love to wear red, and my friends think that I need to spend more time in the sun.