First Steps in Embedded AI on NVIDIA Jetson

The only course in the catalog where AI leaves the screen to run on real hardware, next to a real camera. This is NVIDIA's signature training track: you design, deploy, and defend a complete embedded computer-vision system — from picking the right hardware (the Jetson family as it stands in 2026, from the legacy Nano to the Orin generation) to a fully instrumented system, accelerated with TensorRT and INT8 quantization, hardened to run unattended for years, and compliant with the expectations of the French data-protection authority (CNIL). Eighteen lessons, an end-to-end guided capstone, and one rule with no exceptions: every performance claim is measured in steady state, with the case closed. These are among the highest-paid skills on the edge AI market — equivalent in-person training runs into thousands of euros a day. You walk away with an interview-ready, demonstrable project: an augmented camera that detects, counts, and tracks objects in real time, complete with its GDPR compliance file.