Lead AI Engineer

Job Title: Lead AI Engineer (SC Cleared) Location: Remote (with ad-hoc travel to client’s office in London) Contract Length: 6 Months (with scope to extend) Start Date: ASAP IR35: Inside Interview Process: 1 Stage, MS Teams Clearance Required: SC (needs to have been used within the past 10 months) We are supporting a government client in hiring a Lead AI Engineer to drive AI adoption and governance across a large, security-sensitive engineering organisation. This is a senior technical influence role, pairing hands-on AI engineering practice with the ability to set standards and guardrails adopted across multiple delivery teams. The successful candidate will need demonstrable, hands-on practice with agentic AI engineering tooling, including model selection, prompt/harness engineering and agent-tool integration, alongside a genuine track record of getting engineering standards adopted across teams without formal authority. Initial emphasis will be on landscape setting, evaluation and standard design, shifting over time towards a forward deployed model working directly within delivery teams. This role suits an experienced, senior AI/software engineer who has moved deeply into practical AI-enabled engineering, comfortable operating as a Lead-level individual contributor rather than in a conventional people-management capacity, and confident engaging stakeholders across portfolios, professions, assurance functions and external suppliers. Key Responsibilities Lead horizon scanning across the AI tooling landscape, translating emerging capability into practical, prioritised opportunities for portfoliosBuild and sustain an internal practitioner community, running knowledge sharing, drop-in support and champions networksBuild capability through coaching and targeted training so adoption persists without central supportEngage across government and industry to bring proven approaches in, contribute evidence back, and avoid duplication of workEngage suppliers and technology providers to identify reusable practice and reduce fragmented, supplier-specific approachesSurface and broker resolution of recurring adoption blockers across governance, tooling, data, security and commercial routesRun structured proofs of value with defined hypotheses, baselines and exit conditions, reporting candidly on benefit realisationEmbed with delivery teams as a forward deployed engineer, providing hands-on support to prove approaches in real codebases and transitioning ownership to those teamsAdvise where AI adoption should be constrained, paused or stopped based on risk to security, quality or maintainability Essential Skills Hands-on practice with agentic AI engineering, including building and running agent workflows, model selection, prompt/harness engineering and agent-tool integrationExperience defining engineering standards, guardrails or practices that were adopted across multiple teams, with evidence of how adoption was achievedExperience influencing without formal authority across organisational or team boundaries, including prior engagement with suppliers and technology providersKnowledge of applying AI across the Software Development Lifecycle, including test generation, code review augmentation, documentation and legacy comprehensionAI coding and agent tooling exposure, for example Claude Code, GitHub Copilot, Codex, Kiro or MCP server developmentExperience providing coaching, targeted training or running a champions network, to support a self-sustaining community of practice Desirable Skills Software engineering background in large-scale production systems, sufficient to embed within delivery teams and be credible on their stackExperience with model access and platform routes, such as AWS Bedrock, Azure OpenAI, AI/MCP Gateways, model routing, authentication and cost optimisationExperience running structured evaluations of tools or techniques, including defining hypotheses, baselines, success criteria and candid reporting of negative resultsKnowledge of AI evaluation and measurement processes, such as eval harnesses, regression suites for non-deterministic output and LLM-as-judge approachesAwareness of assurance and security context, such as Secure by Design, DPIA and DPA