Computer Vision Engineer — Edge AI (Manufacturing Safety)
Location: Remote
What This Role Is
Build and deploy real-time computer vision systems for factory safety.
You’ll take models from training → edge deployment → live factory use — and make them work reliably 24/7.
What You’ll Do
- Train & fine-tune object detection models (YOLO) for:
- PPE detection
- Zone intrusion
- Vehicle–pedestrian proximity
- Behavioral safety violations
- Integrate pose estimation for ergonomics & fall detection
- Optimize models for edge inference (ONNX, TensorRT on NVIDIA Jetson)
- Build real-time video pipelines (RTSP → inference → event output via MQTT)
- Implement privacy controls (face blurring, anonymization)
- Own dataset lifecycle: collection, labeling, augmentation, versioning
- Deploy on-site, tune per camera/environment, and iterate from real-world feedback
- Own full lifecycle: data → model → deployment → monitoring
Must Have
- Proven deployment of YOLO models to production (not just experiments)
- Experience with NVIDIA Jetson + TensorRT optimization
- Real-time video pipeline experience (RTSP streams, continuous inference)
- Strong Python + PyTorch + OpenCV
- Solid grasp of precision/recall/mAP and debugging model failures
- Experience managing datasets (CVAT, Roboflow, etc.)
- Ability to ship working systems independently
Strong Plus
- Pose estimation (MediaPipe, RTMPose)
- Object tracking (DeepSORT, ByteTrack)
- MQTT / edge-to-cloud / IoT systems
- Manufacturing or industrial environments
- Anomaly detection / segmentation use cases
- Docker for edge deployment
What Matters
You don’t just build models — you make them work for the real world.
Speed, reliability, and iteration matter more than research.
Job Type: Full-time
Pay: ₹500,000.00 - ₹1,000,000.00 per year
Experience:
- Computer vision: 5 years (Preferred)
- object detection model: 3 years (Preferred)
Work Location: Remote
