Eugene Bagdasarian

Eugene Bagdasarian is an Assistant Professor in the Department of Computer Science at the University of Massachusetts Amherst. His research focuses on security and privacy in AI systems, particularly adversarial machine learning, federated learning vulnerabilities, and contextual integrity frameworks. He holds a PhD from Cornell Tech and a prior engineering degree from Bauman University, with industry experience at Cisco as a software engineer.

Research interests include backdoor attacks in federated learning (developing frameworks like Backdoors101 and Mithridates), privacy-preserving technologies such as AirGapAgent and Ancile, and mitigating vulnerabilities in multi-modal systems. His work has addressed instruction injections, adversarial illusions, and bias in generative models, with coverage in media like The Economist and VentureBeat.

Notable achievements include the Usenix Security Distinguished Paper Award, Apple Scholars in AI/ML Fellowship, and Digital Life Initiative Fellowship. His contributions span technical domains like differential privacy, secure aggregation, and policy-based federated learning frameworks.

Advising and grants: No specific advising/grants listed. However, his research collaborations include developing systems like Ancile (use-based privacy enforcement) and Mithridates (backdoor auditing). Current work trends focus on advancing security in LLM agents, dynamic network firewalls, and ethical AI governance.