Running vLLM inference on NVIDIA Jetson AGX Orin for local, private AI processing without cloud dependencies.