Edge AI Systems: Hardware-Software Co-Design for Intelligent Embedded Platforms addresses one of the most demanding challenges in modern embedded computing: how to build intelligent systems that run AI workloads directly on constrained devices, sensors, microcontrollers, edge gateways, and embedded processors — where power, memory, thermal budgets, and real-time responsiveness cannot be treated as afterthoughts.

For years, AI development and embedded systems engineering evolved largely as separate disciplines. Machine learning models were designed with abundant compute and memory in mind, then handed off to hardware teams to somehow make them fit. Embedded engineers, in turn, optimized hardware without deep visibility into the models it would eventually need to run. That separation no longer works. As AI moves from the cloud to the edge — into autonomous devices, industrial sensors, wearables, robotics, and real-time control systems — the boundary between hardware design and software/model design has to dissolve. Decisions about model architecture, quantization, memory layout, instruction sets, accelerators, and power management are no longer independent; they are deeply, unavoidably intertwined.