This document discusses using mobile edge computing (MEC) at the edge of 5G networks to power smart city surveillance with AI video analytics. It describes how a MEC host running Radisys software can run AI-based video analytics applications at the edge to do real-time facial detection on surveillance camera video streams. Only video streams that match faces would be sent to the surveillance command center, reducing latency, backhaul costs, and improving surveillance accuracy and productivity. The Radisys MEC platform is based on off-the-shelf processors and is open, standards-compliant, and pre-configured to enable and manage edge applications.