Zedong Zhou

Zedong Zhou

Ph.D. Student in Computer Science | Systems Security & AI Agent Security

Zedong Zhou

I am Zedong Zhou, a Ph.D. student in Computer Science at the University of Nevada, Reno (UNR), advised by Dr. Niusen Chen. My research interests lie at the intersection of systems security, operating systems, storage systems, and the security implications of increasingly autonomous AI agents.

I am particularly interested in understanding how privileged software interacts with traditional security boundaries, and how data can remain protected even when the host environment is not fully trusted.

Education

University of Nevada, Reno

Ph.D. in Computer Science
Reno, Nevada, USA
September 2026 – Present

Anhui Polytechnic University

Bachelor’s Degree in Electrical Engineering and Automation
Wuhu, Anhui, China
September 2021 – June 20, 2025

Undergraduate Research Background

My undergraduate research focused on applied intelligent algorithms, with work in visual–inertial SLAM and mapping, battery health prediction, and energy forecasting. I also developed familiarity with computer vision, 3D reconstruction, and the concepts and applications of large language models (LLMs). These interests complement my current Ph.D. research in systems and AI agent security.

Learn more about my research

Research Interests

  • AI Agent Security: Security risks posed by AI agents with elevated permissions, including unauthorized file modification or deletion.
  • Operating Systems and Kernel Security: Linux security mechanisms, access control, and kernel-level enforcement.
  • Storage and Hardware Security: Data integrity and protection mechanisms beyond the host software stack.
  • Security Analysis with AI: Applications of machine learning and program analysis to software security.

Selected Research Project

When AI Agents Go Rogue: Who Protects Your Files?

This research investigates whether host-controlled protection mechanisms can safeguard sensitive files when an AI agent has root privileges. The study examines defenses at different levels of the software stack, including AppArmor, eBPF LSM, and block-layer enforcement, to identify the limits of host-side protection and motivate stronger data-integrity boundaries.

Research/poster project; a formal publication reference will be added when available.

See the Research page for more information about my interests and projects.

Publications

My publications and preprints cover efficient 3D reconstruction, battery health prediction, visual–inertial SLAM, and electricity demand forecasting. See the Publications page for full citations and links to papers and code.

Contact