Computer Vision Engineer

You’ll build the vision systems that our industrial clients run on their factory floors, construction sites and warehouses. That means detection, segmentation, tracking and pose models that keep working in poor lighting, with occlusion and odd camera angles. You’ll take a model from dataset review all the way to a deployed system the client’s operators use every shift.

• Train and fine-tune detection, segmentation, tracking and classification models for real client problems
• Review datasets, write annotation guidelines and find the labeling errors that hurt accuracy
• Evaluate models with the right metrics (mAP, IoU, precision and recall per class) and explain the results to clients in plain terms
• Build video analytics pipelines that process multiple camera streams in real time
• Work with the edge AI team to optimise and deploy your models on target hardware
• Write clean, documented code that another engineer can pick up and maintain

• 2+ years training and shipping computer vision models in production, beyond academic projects
• Strong PyTorch and OpenCV skills
• Hands-on experience with modern detectors such as the YOLO family, RT-DETR or Detectron2
• Solid understanding of CNN and transformer architectures, augmentation and loss functions
• Experience processing video streams, including RTSP feeds and frame sampling

• Experience with NVIDIA DeepStream or GStreamer pipelines
• Exposure to 3D vision, depth cameras or stereo setups
• Experience with multi-object tracking such as ByteTrack or DeepSORT
• Published work, Kaggle results or open-source contributions in vision