Machine Learning Engineer — LLMs, Retrieval and MLOps for NATO with security clearance

ppA multinational defence organisation is building its own applied AI capability, and is looking for an engineer who can take a model all the way into service. You would work on real datasets and real problems, own the pipeline around the model as much as the model itself, and do it inside an environment where reliability and traceability matter more than novelty. /p h3What you would be doing /h3 ul liBringing machine learning and data science methods to new datasets and new questions, and judging honestly how well the result performs and how good the data underneath it is. /li liFinding what is wrong in models, pipelines and datasets, and fixing it in ways that survive contact with production. /li liDesigning, writing, testing, documenting and refactoring the programs, scripts and AI components the capability is made of. /li liWorking to the engineering standards and secure development practices of the organisation, so that what you build can be maintained by someone else. /li liSupporting the whole lifecycle: gathering what is actually needed, choosing how the team works, and automating build, test, release and monitoring. /li liDefining AI modules for integration, producing the build definitions and validating finished modules against agreed functional, quality, security and performance criteria. /li liBuilding and improving the data pipelines that feed all of it, including extraction, transformation and loading work. /li liKeeping colleagues informed — progress, risks and blockers — and sharing delivery ownership through reviews rather than handovers. /li liWatching what is arriving in the field and contributing to technology assessments, roadmaps and internal knowledge sharing. /li /ul h3What you would bring /h3 ul liHands-on history of developing, optimising, deploying and maintaining complete AI pipelines, including training, packaging, monitoring and lifecycle management. /li liStrong programming alongside the machine learning: software engineering discipline applied to applied AI work. /li liA solid grasp of model evaluation — how performance is measured, how it is assessed and how a model is actually improved. /li liPractical use of pre-trained and foundation models, large language models and generative techniques on problems that needed solving. /li liRetrieval-augmented generation, embeddings, vector stores and production agent backends, with frameworks such as LangChain, LlamaIndex or Pydantic AI. /li liMLOps in earnest: version control, continuous integration and delivery, experiment and model lifecycle practice, automated build and release. /li liBackend craft — REST services and modern Python with FastAPI, Pydantic or similar. /li liContainers and orchestration: Docker, Kubernetes, Helm, cloud provisioning, and workflow orchestration such as Airflow or Argo. /li liGuardrails and operational control for language-model systems: observability, logging and monitoring that tell you when something has drifted. /li liSQL and NoSQL databases, and enough TypeScript, Node.js or Next.js to meet the front end halfway. /li liNice to have: experience of working in secure, restricted or disconnected environments. /li /ul pExtensions are offered where the work goes well. Applications are reviewed as they arrive. /p /p #J-18808-Ljbffr