April 30, 2026
Most software we rely on must be certified safe without them having to expose their source code. Software across various domains, like healthcare, finance, and autonomous vehicles, etc., need to ensure functional correctness and the correct usage of vast scales of sensitive data. Presently, systems use ad-hoc safety checks in software, but this enforcement needs to be streamlined, automated and be non-invasive. So in this talk, I’ll present our automaton-based distributed monitoring framework comprising: (a) an expressive specification language for control and data-aware policies, (b) a (visibly pushdown) automaton-based enforcement mechanism, and (c) a distributed monitor that does not require any invasive changes to the source code.
This talk offers insights for those seeking to learn about: (a) an application of visibly pushdown automata to runtime monitoring of tree-based control-flow properties, (b) a scalable distributed monitor implementation, or (c) how a network layer can be repurposed for monitoring blackbox systems.
(This is based on SafeTree, my OOPSLA’25 paper jointly done with Brighten Godfrey (UIUC) and Justin Hsu.)
About Karuna Grewal
Karuna Grewal is a Ph.D. candidate in the Computer Science department at Cornell University advised by Prof. Justin Hsu on language-based security and runtime monitoring mechanisms for systems with opaque behavior that cannot be statically verified. Before starting her Ph.D., she worked on software-defined networking at Microsoft and on a theory-oriented real-time operating system at MPI SWS (during her undergrad at BITS, Pilani in India).