Week 17: AI at the Edge

Where inference runs changes the system boundary.

A model can execute on a local device, on a nearby edge computer, in a remote cloud service, or across a hybrid design. Each placement changes latency, connectivity, privacy exposure, memory limits, power use, update strategy, and failure behavior.

This week treats deployment as an engineering decision rather than a technology preference. In the Edge vs Cloud Deployment Simulator, you will change network and device constraints and defend a placement from evidence.

There is no universal winner. The useful answer is the placement that satisfies the requirements you can actually justify.