Eugene Bagdasarian

Eugene's work focuses on security and privacy in emerging AI-based systems and agentic use-cases under real-life conditions and attacks. He completed his PhD at Cornell Tech where his research was recognized by Apple Scholars in AI/ML and Digital Life Initiative fellowships and Usenix Security Distinguished Paper Award. Prior to joining grad school, he received an engineering degree from Bauman University and worked at Cisco as a software engineer. 

His work on security focused on backdoor attacks in federated learning leading to new frameworks Backdoors101 and Mithridates. One of the proposed backdoor attacks on generative language models adding bias was covered by VentureBeat and The Economist. Eugene's recent work includes studies on vulnerabilities in multi-modal systems: instruction injections, adversarial illusions, and adding biases to text-to-image models.

On the privacy angle, Eugene worked on the AirGapAgent privacy protection for conversational  LLM agents and operationalizing Contextual Integrity. He worked on aspects of differential privacy including fairness trade-offs, applications to location heatmaps, and tokenization methods for private federated learning. Additionally he helped to build the Ancile system that enforces use-based privacy of user data.