About me

Hello! I’m Aaryan (Pronouns: He/Him), a Computer Science Ph.D. student at the School of Computer, Information, and Data Sciences in the madS&P and madPL groups at The University of Wisconsin-Madison. I am broadly interested in privacy and programming languages. I am currently doing an independent study with Prof. Ethan Cecchetti. Before coming to Madison, I did my bachelor’s at Penn State University in Computer science with honors in computer science, where I was advised by Prof. Danfeng Zhang.

Research

I am broadly interested in privacy, programming languages, and program analysis. My current project focuses on using a security type system to enforce information flow control in Android platform/Java code. We are building a pluggable type system for the Java compiler which enforces noninterference. Our information-flow labels are privacy ‘nutrition’ labels a representation of privacy policies which summarizes the key policies and is easier for users to interpret - just like nutrition labels. Privacy-labels are adopted by Google’s Play Store as the data safety section and we want to connect them to the source code using information flow control.

During undergrad, I worked on developing a language and novel type system for automatically enforcing approximate differentially privacy for programs that use Gaussian noise. The type system is sound and transforms a probabilistic differentially private program into a deterministic one which makes the privacy costs explicit using the characteristic function of privacy loss random variable. This deterministic program can be verified using an existing off the shelf verifier, like Dafny or CPAChecker.

Another interesting project I worked on was a final project for CS 782 - Advanced Security and Privacy (Taught by Prof. Kaseem Fawaz) where I used Large language models and static call graph analysis for checking consistency between commit messages and code changes on GitHub. I have also done an independent study with Prof. Loris D’Antoni studying capabilities of large language models for program synthesis. I explored grammar-constrained decoding, a technique to generate output constrained by context-free grammar.

Personal

Outside of work, I like weightlifting, running, swimming, and reading books. I am also getting into tennis lately, so if you know how to play let’s hit the court.

News

January 2025: I started an independent study with Prof. Ethan Checcheti

November 2024: I attended Midwest Security Workshop

June 2024: I attended Oregon Programming Languages Summer School

September 2023: Started Ph.D. at UW Madison Computer Sciences

June 2023: I attended my first academic conference - Programming language Design and Implementation