Join us!

About the role

Most organisations have now run their AI experiments. Far fewer have got anything into production that holds up under real load, real data and real scrutiny.

You would be our most experienced AI practitioner in the UK, working closely with engineering teams and senior client stakeholders. The role has four parts:

• Technical leadership. Setting the architecture and standards for how we build AI systems, and staying hands-on.
• Consulting. Working with clients to turn AI ambition into a plan that holds up, including advising against approaches that will not.
• New business. Identifying and shaping opportunities, and leading the technical side of proposals and pitches.
• Growing the team. Mentoring, training and hiring, so that AI capability is shared across the team rather than held by one person.

What you’ll be responsible for

• Setting the technical direction for AI and data work across Futurice UK, and being responsible for the quality of what we build
• Architecting and building production AI systems: agentic applications, retrieval and knowledge systems, applied ML, and the data platforms beneath them
• Establishing how we evaluate AI systems, so that quality, cost, latency and risk are measured rather than assumed
• Advising client leadership on AI strategy, operating model, build-versus-buy, and moving from pilot to production
• Designing governance, privacy, security and responsible AI practice into the work from the start
• Leading the technical shaping of new opportunities: qualification, solution design, estimation and pitches
• Contributing to our profile through writing and speaking, and growing UK capability through mentoring, training and hiring
• Working with the wider Futurice AI community across Europe

Skills & Requirements

We’d love to hear from you if you have:

Technical depth

• Production AI systems. Experience taking LLM-based and agentic systems into production and supporting them afterwards, and a clear view of what separates a working demo from a production system.
• Agentic architecture. Tool use, planning and execution loops, multi-agent orchestration, memory and state, human-in-the-loop checkpoints and failure recovery, with attention to the surrounding system rather than the model alone.
• Context engineering. Retrieval strategy, chunking and compaction, caching, structured inputs, freshness, and the associated cost trade-offs.
• Evaluation and observability. Eval design, systematic error analysis, the limits of LLM-as-judge, regression suites for non-deterministic systems, and tracing agent behaviour in production. We treat this as central.
• Retrieval and knowledge systems. Hybrid search, re-ranking, evaluating retrieval quality in its own right, graph-based approaches, and judgement about when retrieval is not the right approach.
• Model selection and adaptation. Choosing between frontier, small and open-weight models, or none at all. Fine-tuning, distillation and RL-based post-training, and a sense of when none of them are needed.
• Classical ML and data science. Forecasting, optimisation, causal inference, recommendation and anomaly detection. A good deal of useful work is not generative.
• Data and platform foundations. Pipelines and orchestration, lakehouse and streaming patterns, data contracts and quality, semantic layers and lineage. Data foundations often set the limit on what a client can do with AI.
• Cloud and operations. Hands-on experience with AWS, Azure or GCP
• AI security. Prompt and indirect injection, tool-use and permission design, data exfiltration paths, sandboxing, agent identity and least privilege, and red-teaming.
• Software engineering. Strong Python and SQL, sound design instincts, testing discipline, code review, and comfort working in an existing codebase.
• AI-assisted engineering. Working knowledge of coding agents and AI tooling, where they help and where they add risk, and the ability to support client teams adopting them