AI Researcher

Description

The company is a global technology leader with over 83,000 employees across five continents, including more than 7,500 in the UK. Its work covers defence, space, aerospace and digital security, supporting national security and stability.

You will join the software, AI and research group in the UK and work alongside AI engineers, assurance specialists, human-machine teaming researchers, product owners, domain experts, and data and software engineers.

The aim is to turn hard operational and business problems into tested research results, proofs of concept and reusable technical capability. This work helps win bids, improves the company's offers and leads to better outcomes for customers.

Your outputs will feed a shared internal catalogue of AI capabilities and help teams adopt AI responsibly, faster. You will link with data, digital, research and engineering specialists across the UK business and the wider group, develop emerging AI technologies for future use, and act as a technical expert on advanced AI.

The role suits someone who can do research and also turn it into a product. Having supported or led bids is highly advantageous.

Responsibilities

• Run applied and experimental AI research on complex customer and business problems in defence, aerospace, cyber security, rail, critical national infrastructure and related fields.
• Build state-of-the-art AI and ML solutions, proofs of concept and research prototypes using real data and problems framed around operational needs.
• Investigate, design, build and assess advanced methods such as deep learning, multimodal AI, self-supervised learning, foundation models, generative AI, computer vision, NLP and LLMs, time-series analytics and, where relevant, reinforcement learning.
• Turn business and customer needs into clear research questions, experiment plans, technical requirements and measurable success criteria.
• Design strong evaluation approaches using baselines, benchmarks, ablation studies, uncertainty estimates, robustness tests and metrics that matter operationally.
• Work with the AI V&V, AI assurance, human-machine teaming and applied AI groups so that outputs are trustworthy, human-centred, explainable and fit for future operational use.
• Set up reproducible pipelines covering data preparation, feature engineering, training, experiment tracking, evaluation and reporting.
• Build responsible AI into every stage of research, covering robustness, safety, explainability, transparency, fairness, privacy, security, and alignment with MOD, regulatory and company governance.
• Produce proofs of concept, publications, invention disclosures, patents, technical reports and reusable assets on topics such as multimodal learning, self-supervised learning, foundation models and human-AI collaboration.
• Prepare research outputs for handover to AI engineering teams, including demonstrator code, model cards, experiment reports, design notes and handover documents.
• Support bids, proofs of concept, demos, customer workshops, innovation campaigns and briefings by explaining research to technical and non-technical audiences.
• Work with data engineers, architects and domain experts on data acquisition, labelling, synthetic data, integration of third-party data and data quality.
• Track major AI research and technology trends, judge their relevance to the company's markets, run trials and share good practice to speed up responsible adoption.

Requirements

• A degree or master's, or an equivalent in a relevant software or AI subject, or equivalent experience. Relevant areas include AI, machine learning, computer science, data science, mathematics, engineering, physics or a closely related technical field.
• Strong Python skills and solid practice in modern software engineering and research, including testing, code quality, reproducibility and team development.
• Experience doing AI/ML research or advanced AI development in complex technical settings, ideally in defence, aviation, rail, cyber security, safety-critical, mission-critical or similarly regulated domains.
• Expertise in ML and DL methods for supervised, unsupervised and self-supervised learning and, where relevant, reinforcement learning.
• A track record of taking AI research from problem framing through experiment design, model development, evaluation and prototype demonstration.
• Hands-on work in at least one advanced AI area, such as deep neural networks, computer vision, NLP/LLMs, multimodal AI, self-supervised learning, reinforcement learning, time-series analytics or foundation models.
• Experience with PyTorch, TensorFlow, scikit-learn and Hugging Face Transformers, plus OpenCV for vision work.
• Strong grasp of experimental design, statistical evaluation, benchmarking and model validation.
• Experience with experiment tracking and reproducibility tools such as MLflow, Weights & Biases or equivalent.
• Data wrangling and analysis with Pandas, NumPy and SQL.
• Knowledge of model optimisation and moving research into prototypes, including ONNX, TorchScript, FastAPI/gRPC and GPU acceleration.
• Awareness of responsible AI and security, including explainability, privacy-preserving methods, bias assessment, safety, assurance, adversarial robustness and secure AI.
• Strong scientific and technical writing, including technical reports, papers, invention disclosures, model cards and experiment reports.
• A proven record of producing high-quality technical documentation, research outputs, model evaluations or stakeholder briefings.
• Ability to engage and influence varied stakeholders, including product engineering leaders, customers, design authorities, project management, IS/IT, research partners and academics.
• Effective in a matrix organisation: collaborative, drives outcomes and keeps scientific rigour.
• Clear communicator who can explain complex AI concepts to technical and non-technical people.
• Curious, creative and rigorous, able to challenge constructively and find new solutions to unclear problems.
• Open and encouraging, welcoming idea sharing and technical debate.
• Desirable: PhD or equivalent research experience in AI, machine learning, computer science, mathematics, engineering, physics or a related field.
• Desirable: peer-reviewed publications, patents, invention disclosures, open-source contributions or recognised technical innovations.
• Desirable: experience governing architecture, research designs or detailed technical designs across a project lifecycle.
• Desirable: large-scale data initiatives, labelling strategies, synthetic data generation and data quality management.
• Desirable: MLOps practice and cloud platforms for AI deployment.
• Desirable: working with academic partners, research institutions, grant-funded programmes or collaborative consortia.
• Desirable: contributing to bids, proposals, customer demonstrations or externally funded innovation.
• Desirable: cloud AI services, HPC, containerisation and secure research environments.
• Desirable: Spark or similar, and CUDA basics.

Conditions

Based in Reading (RG2 6GF) with hybrid working.

You must be a sole British national and hold Security Clearance (SC) without any caveats. Holding SC already is advantageous. If you do not hold it, you must undergo, obtain and maintain SC before starting. Guidance is on the UK Security Vetting pages of GOV.UK.

For full SC you generally need to have lived in the UK for the last 5 years. In some cases, 3 years of UK residence over the last 5 years may be accepted, with extra overseas checks.

The company is an equal opportunities employer.

Benefits

• An annual bonus, a strong pension contribution, private healthcare and development-focused targets.
• Market-leading training and development, plus mentoring and coaching from leading specialists in the industry.
• Support for academic study and professional memberships.
• Relocation support where required.
• Private medical insurance, the option to buy or sell annual leave, a cycle to work scheme, employee discounts, a paid volunteering day and stocks and shares, depending on the role.
• The company hires internally first, so you can move into other domains, products, sites or countries, and choose a technical or leadership path.