The emergence of edge AI systems—AI deployed on resource-constrained, often battery-powered, devices at the edge of the network—presents critical security challenges. These systems are increasingly vulnerable to hardware-level threats, including side-channel attacks, fault injections, etc., particularly when optimized for performance.

This Research Fellow position focuses on AI security in the context of hardware-constrained edge devices, investigating how hardware acceleration can be leveraged by adversaries to compromise AI systems' robustness. The role involves designing secure AI accelerators, analyzing attack surfaces introduced by approximation, and developing a performance-security trade-off framework to guide secure AIoT deployment.

About the person: