Enterprise AI security
Defense, evaluation, and observability
Research focus
Distributed systems
Data, cloud, and secure containers
Machine learning
Detection models and production ML
Agent infrastructure
MCP and tool-using agent security
Featured system
ADR — Agentic Detection System
Open-source security observability, benchmarking, and threat detection for enterprise agents.
ADR / Detection pipeline
Agent trace→Policy checks→Risk score
Prompt-injection pattern detected · high severity
Engineering leadership
Uber
Senior Engineering ManagerData infrastructure · ML · AI safety
Alibaba Cloud
Senior Engineering Manager / Staff EngineerCloud and infrastructure systems
Stanford
Postdoctoral Researcher with Sachin Katti (VP of OpenAI)Networked systems
Selected publications
Google Scholar ↗ADR: Agentic Detection for Enterprise AI Security
Li, Hu, Xu, et al., Pengyu Zhang, et al.
FastIOV: Passthrough Network I/O Virtualization
Liu, Guo, Jiang, Song, Pengyu Zhang, et al.
Network Startup for Secure Containers
Liu, Guo, Jiang, Pengyu Zhang, et al.
Recognition and service
Awards
- ACM SIGMOBILE Doctoral Dissertation Award
- UMass Outstanding Dissertation Award
- MobiCom Best Paper Runner-Up
- UbiComp Honorable Mention
- MobiSys and SenSys Best Paper nominees
Program committees
- NSDI · 2027
- MobiCom · 2019, 2020, 2021
- MobiSys · 2017, 2019
- CoNEXT · 2018
I received my Ph.D. from UMass Amherst, advised by Deepak Ganesan, and was a postdoctoral researcher at Stanford with Sachin Katti. I hold bachelor’s and master’s degrees from Tsinghua University.