This course explores the complete Edge AI deployment pipeline, guiding students through the process of training convolutional neural networks (CNNs) on GPU-based platforms and accelerating inference on FPGA-based hardware. Students will develop, train, and evaluate CNN models; convert models using ONNX-based workflows; perform quantization and optimization; and deploy designs onto FPGA accelerator architectures. Emphasis is placed on hardware-software co-design, performance-per-watt analysis, resource utilization tradeoffs, and real-time inference constraints. A semester-long applied project integrates machine learning theory with digital hardware implementation.