엣지 AI 및 배포
TensorRT, ONNX, quantization — making learned models fast and lean enough to run on constrained hardware at the edge.