Computer Vision Engineer

Duration: 6+Months Role Overview You will take computer vision models from development to real-time production on NVIDIA Jetson Orin edge devices, supporting process control, safety, and quality use cases in industrial plant operations. On-device performance, latency, and long-run stability matter as much as model accuracy. Key Responsibilities Edge Model Optimization & Deployment

• Convert and optimize PyTorch vision models (detection, segmentation, classification) to ONNX and TensorRT for Jetson Orin
• Package inference workloads as Docker containers on JetPack/L4T (ARM64) with reproducible builds and deployments
• Work with data scientists to make models edge-ready (input sizing, architecture trade-offs, pre/post-processing) Real-Time Video Pipelines
• Build and tune low-latency RTSP/GStreamer pipelines using hardware decode and zero-copy (NVMM) memory
• Develop DeepStream pipelines, including custom C++ parsers/plugins for non-standard model outputs
• Write performance-critical components in C++ and CUDA; use Python for integration and tooling
• Deliver model outputs to downstream systems (alerts, dashboards, PLC/control-system interfaces) Validation, Profiling & Production Support
• Profile latency, throughput, GPU/CPU/memory utilization, and thermals (Nsight Systems, tegrastats) and remove bottlenecks
• Engineer for 24/7 stability: stream reconnection, watchdogs, backpressure, frame-drop control, logging, and health monitoring
• Remotely support field testing and troubleshooting of deployed systems
• Document deployment procedures, configurations, and runbooks Must Have
• Computer vision / ML engineering experience with models delivered to edge or embedded systems in production
• Hands-on NVIDIA Jetson Orin (AGX Orin / Orin NX / Orin Nano)
• Model optimization and deployment with TensorRT and ONNX
• Python and PyTorch
• Docker on ARM64 / JetPack / L4T
• Profiling and tuning for latency and long-running stability on constrained hardware
• Strong communication skills for working directly with client operations, automation, and IT/OT teams Good to Have
• NVIDIA DeepStream SDK, including custom plugins or output parsers
• CUDA and C++
• RTSP/GStreamer pipelines with IP cameras
• Industrial domain experience (mining, chemicals, manufacturing, energy) and OT integration (PLC, OPC UA, Modbus)
• Edge hardware and camera selection for remote sites
• Fleet management, remote updates, or monitoring for multiple edge devices
• Experience with enterprise change control and cybersecurity review processes Key Skills: Computer Vision | NVIDIA Jetson | TensorRT | ONNX | DeepStream | CUDA | C++ | Python | PyTorch | GStreamer | Docker | Edge Deployment