Computer Vision Engineer

Role overview Build and maintain computer vision models that detect and classify defects in industrial inspection imagery. The work spans model training, data quality, edge deployment, and evaluation, with results used to guide what operators inspect and act on.

Responsibilities - Train and validate object detection and classification models for structural, surface, and component defects. - Maintain data pipelines for annotation, dataset versioning, and augmentation. - Address domain shift and class imbalance in limited industrial inspection datasets. - Optimise and package models for constrained edge hardware. - Work with robotics engineers on image capture settings that affect model performance. - Create evaluation frameworks and regression checks across deployment environments.

Requirements - At least 3 years of production computer vision engineering experience. - Strong Python skills and proficiency with PyTorch or a comparable deep learning framework. - Experience with object detection and segmentation architectures such as YOLO-family models or Detectron2. - Knowledge of domain adaptation, fine-tuning, and transfer learning for industrial data. - Experience deploying models to edge devices using tools such as TensorRT or ONNX. - Familiarity with active learning, annotation tools, and data quality practices.

Benefits and work setup Flexible or remote work is available within Australia, with office options in Melbourne and Sydney. Periodic travel to infrastructure, rail, and construction sites is part of the role.