Company Profile
Location: Waterbeach, UK
Department: Engineering
Reports to: Chief Scientist
Type: Full-time, Permanent
Company Profile
Thermoteknix Systems Ltd, founded in 1982 and based in Cambridge, is a world leader in thermal imaging, night vision, and augmented reality systems. Our award-winning technologies are trusted by defence, security, and industrial customers in over 70 countries. With in-house R&D, cutting-edge innovation, and a global reputation for quality, we design and manufacture solutions that perform in the toughest environments.
About You
Reporting to the Chief Scientist, you will research, develop and validate AI, machine learning and image processing algorithms for our thermal imaging cores and wider imaging product range. AI will form a substantial part of the role, with responsibility for developing and training models alongside conventional image processing, detector correction and calibration.
You will take ideas from scientific investigation and proof of concept through to validated product implementation. Working with software, firmware, FPGA, electronics and optical engineering colleagues, you will balance image quality and model performance with real-time processing, memory, latency and power constraints.
Key Responsibilities
AI and Machine Learning Develop and train machine learning and deep learning models for imaging applications, including image enhancement, object detection, classification, segmentation and scene understanding.
Select and evaluate model architectures, training methods and loss functions, comparing results against conventional algorithms and defined product requirements.
Investigate AI methods for noise reduction, image restoration, super-resolution and image fusion, assessing image fidelity and behaviour on unfamiliar scenes.
Review emerging AI research, reproduce promising methods and assess their feasibility for Thermoteknix products.
Image Processing and Imaging Science Develop conventional algorithms for noise reduction, contrast enhancement, dynamic range compression, detail enhancement, registration and stabilisation.
Develop detector correction and calibration methods, including non-uniformity correction, defective pixel compensation, temperature drift compensation and shutterless processing.
Analyse image artefacts and system limitations, investigating the effects of detectors, optics, electronics and processing software, and combining analytical and learned methods where beneficial.
Data and Validation Plan image and video data collection and annotation with engineering colleagues, and maintain representative datasets with clear provenance and version control.
Design reproducible experiments and separate training, validation and test datasets to prevent data leakage; assess generalisation across cameras, environments and operating conditions.
Define image quality and AI performance metrics, including accuracy, false alarms, missed detections, latency and temporal stability, as relevant to the application.
Validate algorithms through laboratory testing and field evaluation, investigate failure cases and ensure enhancement preserves scene content and avoids misleading artefacts.
Embedded Implementation and Product Support Prototype algorithms and models in Python, MATLAB or similar tools, and work with engineering teams to integrate them into real-time product software and processing pipelines.
Optimise models for embedded deployment using appropriate methods such as quantisation, pruning and hardware acceleration, measuring effects on accuracy, image quality and resource use.
Support the investigation of production and customer image quality issues, using data analysis and experiments to recommend and validate improvements.
Research and Technical Collaboration Work with the Chief Scientist to define research priorities, assess technical risks and contribute to product feasibility studies, design reviews and intellectual property…