About

I am a PhD candidate in Computer Science at UMass Amherst, advised by Amir Houmansadr and co-advised by Eugene Bagdasarian.

My research centers on security, privacy, and behavioral properties of AI agent systems. I study how LLM agents interact with tools and connected software, and how that perception-action loop creates new risks such as prompt injection, information leakage, behavioral manipulation, and agent fingerprinting.

My goal is to adapt security thinking for autonomous AI systems: identifying agent-specific vulnerabilities and designing defenses that make agents safer to deploy in real-world environments.

---

🔬 Research Interests

• AI Agent Security – Understanding vulnerabilities specific to AI agents; designing attacks and defenses
• Web & Browser Agent Security – Dynamic prompt injection, agent observation manipulation, and secure-by-design browser agents
• Agent Privacy & Attribution – Prompt reconstruction and user profiling from encrypted browsing traces; multi-layer agent fingerprinting
• LLM Fairness & Behavior – Bias measurement; behavioral drift and persuasion effects in long-running agents
• Federated Learning & Unlearning – Privacy-preserving algorithms and selective knowledge removal in distributed systems
  
  

---

📢 Recent News