Edge AI Engineer

Many of our clients can’t send video to the cloud, either because of bandwidth, latency or data privacy. So the model has to run on a small device sitting next to the camera. You’ll take trained models and make them run fast and reliably on NVIDIA Jetson and other edge hardware, hitting real-time frame rates within tight power and memory limits.

• Convert and optimise models with TensorRT, ONNX Runtime and quantisation (FP16 and INT8)
• Build real-time inference pipelines on Jetson Orin, Xavier and Nano devices
• Profile CPU, GPU and memory use, then remove the bottlenecks that drop frame rates
• Write performance-critical code in C++ where Python is too slow
• Set up devices for field deployment, including remote updates, logging and recovery after power loss
• Test models on real hardware with real camera feeds before anything ships to a client

• 2+ years deploying ML models on edge or embedded devices
• Strong C++ and Python
• Hands-on experience with TensorRT and ONNX model conversion
• Experience with NVIDIA Jetson, JetPack and CUDA
• Comfort on Linux, including cross-compiling, systemd services and debugging on-device

• Experience with NVIDIA DeepStream or Triton Inference Server
• Exposure to other edge platforms such as Qualcomm SNPE, Intel OpenVINO or Hailo
• Background in embedded systems or electronics
• Experience with depth sensors such as Intel RealSense or Stereolabs ZED