Hire AI evals engineers in the UK — 5-day shortlist, AISI-aware.

Worked example, senior evals engineer into a London-HQ regulated fintech (placed 2026-Q1, FCA-supervised). Target role: UK-local compensation would have been £140K base + 15% overhead = £161K/yr fully loaded. Actual placement: CEE senior evals engineer at €86K/yr on B2B (≈£74K total-comp-equivalent). Recruo fee: 15% of £74K = £11.1K one-time. Classical UK agency alternative, at the conservative 22% end of the 22–25% UK standard: 22% of £140K = £30.8K. Year-1 delta: (£161K + £30.8K) − (£74K + £11.1K) = £106.7K in the client's favour. From year 2 onward the recurring delta holds at roughly £87K because the placement fee does not repeat, and the candidate's evaluation harness keeps compounding in value against the FCA model-risk review cycle.

The UK is the only major Western market where evaluation is now a named, visible, budgeted function inside AI teams — not a subtask buried under ML engineering. The UK AI Safety Institute (AISI) released Inspect AI as an open-source evaluation framework in 2024 and has continued to expand it through 2025–2026, and the framework is now the default starting point for internal evaluation harnesses at UK AI labs, regulated fintechs, and the top tier of UK enterprise. This matters for hiring: when a UK buyer writes 'evals engineer' in a job spec, they increasingly mean 'someone who can extend Inspect AI, stand up task-specific evaluations, and hand those evaluations to a compliance reviewer intact'. The supply of engineers who have done this end-to-end in production is measured in low hundreds across the UK — we see the same twenty or thirty names rotating between DeepMind, Anthropic London, UK AISI itself, and the safety teams inside two or three large consultancies.

The hiring-geography story is tighter than for general LLM engineering. London dominates because the regulated buyers are there; Cambridge and Oxford concentrate the AI-safety-native candidates because of the university research ecosystem and the cluster of DeepMind-adjacent and ex-OpenAI spinouts that settled around Cambridge in 2024–2025. For UK clients outside that golden triangle — a Bristol scale-up, a Manchester insurer, an Edinburgh fintech — hiring a UK-local senior evals engineer means competing head-on with the same London salary bands and frequently losing. Our UK recruitment agency practice exists specifically to let those firms field a credible evals function without relocating to Zone 1.

Evals hiring also carries an unusual compliance-adjacent profile. Evals engineers are the practical interface between an AI engineering team and three UK oversight regimes at once: ICO AI guidance (bias and explainability), UK GDPR (data-protection-by-design on training and eval data), and, for FCA/PRA-regulated firms, the model-risk management framework published in SS1/23. A competent evals engineer turns each of these from a board-level anxiety into a repeatable test that runs on every model iteration. That is why UK enterprise buyers — especially those going through procurement due diligence with a regulated customer — treat the evals hire as schedule-critical, and why the shortlist quality matters more than for almost any other AI hire. Our EU AI Act and UK GDPR readiness checks are run on every candidate in the evals pipeline as a matter of course.