Develop, evaluate and improve Computer Vision and machine-learning solutions using the existing codebase and delivery approach.
Scope of Work
- Train, fine-tune and evaluate Computer Vision models
- Prepare and maintain datasets, experiments, evaluation results and model artefacts
- Investigate false detections, missed events, tracking errors and model degradation
- Identify edge cases, define labelling tasks and review annotation quality
- Analyse conditions such as layout, lighting, camera angle and crowding that may affect performance
- Support model packaging, versioning, deployment, release and rollback activities
- Perform retraining, threshold tuning and regression testing when required
- Maintain model documentation, evaluation reports, limitations and handover materials.
- Monitor model in production and maintain the model performance
- Follow organization's privacy, security, data-retention and AI-governance requirements
Minimum Required Experience and Qualifications
- Practical Python, deep-learning and Computer Vision experience
- Experience with at least one Computer Vision workflow such as detection, classification, tracking, pose estimation or video analytics
- Familiarity with frameworks such as PyTorch, TensorFlow, OpenCV or similar tools
- Understanding of dataset preparation, model training, precision, recall, F1 and latency
- Able to analyse model failures and recommend data, configuration or model improvements.
- Familiarity with annotation tools and labelling quality checks
- Good documentation and communication skills
- Experience with Docker, MLOps, experiment tracking, ONNX, TensorRT or GPU inference is useful but not mandatory
- Hospitality, F&B or camera-based analytics experience is useful but not mandatory