Research

I study the security, safety, and reliability of large language models, with a focus on attention mechanisms and adversarial vulnerabilities.

About Me

I am a 4th-year Ph.D. student in Computer Science at the University of California, Riverside, working with Prof. Nael Abu-Ghazaleh. My research focuses on AI safety, security, and reliability, with a particular emphasis on large language models (LLMs) and multimodal AI systems. I study how architectural components, especially attention mechanisms, can introduce vulnerabilities such as jailbreaks, privacy leakage, and alignment failures. My work has appeared in top venues including EMNLP, ACL, USENIX Security, and IEEE S&P, where I have contributed first-author research on attention-based attacks, membership inference, and adversarial robustness of LLMs.

More broadly, I am interested in building secure, robust, and trustworthy AI systems by combining theory-driven analysis with hands-on experimentation. I have experience developing and evaluating machine learning models across NLP, vision, and multimodal settings using PyTorch and HuggingFace, and I actively explore real-world attack surfaces and defenses in deployed systems. Prior to my Ph.D., I earned a B.Sc. in Electrical Engineering from Sharif University of Technology, where I built a strong foundation in machine learning, signal processing, and systems. I am always excited to collaborate on research at the intersection of AI safety, adversarial ML, and applied LLM systems.

News ⬇️