Computer Vision Engineer, AtkinsRéalis

• Proficient in Python and / or C++ with hands on experience with tools such as PyTorch and OpenCV
• Understanding and proven experience utilising a wide range of computer vision algorithms including:
• Classical deterministic image processing
• 2D based deep learning including object detection and similar processes
• 3D scene reconstruction and spatial reasoning
• Image translation alignment and warping
• Experience managing the full model lifecycle: training, validation, tuning, deployment, maintenance and iterative improvement
• Understanding of hardware acceleration for development and deployment, including technologies such as CUDA, DirectML, and ONNX execution providers
• Solid experience using Git (or equivalent) and working within CI/CD pipelines
• Ability to produce clear, structured technical documentation, including design decisions, data and model lineage, evaluation results, and operational guidance, suitable for both engineering and non‑specialist stakeholders
• Experience using open‑source data labelling tools (e.g. Label Studio or similar)
• Experience integrating multi‑modal vision data, including:
• Standard 2D imagery (colour and grayscale)
• Depth sensing technologies
• Hyperspectral imagery
• X‑ray and gamma imaging

Desirable

• As a Computer Vision Engineer, you will contribute to the design, development, and deployment of production‑grade computer vision systems, supporting both intelligence generation and autonomous action
• You will work in multidisciplinary teams alongside engineers, consultants, and domain experts, transforming complex technical problems into robust, reusable, and maintainable solutions with real‑world impact
• You will be involved across the full computer vision lifecycle, from problem definition through to deployment and continuous improvement
• Problem framing & solution design: Translate product, robotics, or inspection needs into well‑scoped computer vision tasks and end‑to‑end processing pipelines
• Data pipelines: Plan and execute data collection, curation, augmentation, and annotation strategies for computer vision datasets
• Modelling & algorithms: Implement, adapt, and fine‑tune computer vision models and methodologies, spanning both 2D and 3D techniques
• Model & pipeline evaluation: Define and implement appropriate performance metrics (accuracy, robustness, latency, efficiency) and carry out structured error and failure‑mode analysis
• Optimisation & deployment: Optimise models and pipelines for deployment on edge and accelerated platforms, balancing accuracy, latency, resource usage, and reliability
• Integration & productionisation: Build and maintain production‑grade inference and processing services suitable for diverse deployment environments. Use version control, CI/CD, and dataset/model versioning to support reproducible and maintainable delivery

• This role may require security clearance and offers of employment will be dependent on obtaining the relevant level of clearance. If this is necessary, it will be discussed with you at interview
• The vetting process is delivered by United Kingdom Security Vetting (UKSV) and may require candidates to provide proof of residency in the UK of 5 years or longer
• If applying to this role please do not make reference to (in conversation) or include in your application or CV, details of any current or previously held security clearance