How can we build large applications with strong security and privacy properties? How can we balance complexity of the application's logic with the complexity of the cryptographic tools needed to achieve security?
 In this talk, I will discuss an answer to this question: the **programmable cryptography stack**. A programmable cryptosystem allows a developer to build applications that securely compute over encrypted, secret, or distributed data. The programmable cryptography stack allows developers to express these secure applications using high level abstractions. 

I will discuss three of my contributions to the programmable cryptography stack. First, a new programmable cryptosystem with new security properties. Second, a compiler infrastructure that multiplexes between different high-level programming languages and different programmable cryptosystems. Third, an SMT solver suitable for verifying programmable cryptography.
 These ideas have been used in industry to find real bugs and verify real programmable cryptosystems. They are also the basis for ongoing industrial implementations of private exchanges and private outsourcing.

Biography

Alex Ozdemir is a PhD student at Stanford advised by Dan Boneh and Clark Barrett. His work combines ideas from cryptography, compilers, and verification. He and his collaborators have published 18 papers at top-tier venues in computer security, verification, and distributed systems, and they won the Best Tool Paper award at TACAS 2022.