Hire LLM engineers in the UK — 5-day shortlist, IR35-safe.

Worked example, senior LLM engineer into a Series B London fintech (placed 2025-Q4). Target role: UK-local compensation would have been £135K base + 15% NI / pension / equipment overhead = £155K/yr fully loaded. Actual placement: CEE senior LLM engineer at €78K/yr on B2B contract (≈£67K). Recruo fee: 15% of the £67K base = £10K one-time. Classical UK agency alternative fee: 22% of a UK-local £135K = £29.7K. Year-1 total delta: (£155K + £29.7K) − (£67K + £10K + EOR cost £0) = £107.7K in the client's favour. This compresses to roughly £85K delta annually from year 2 onwards when the agency fee is no longer recurring.

London's senior-AI hiring market compressed in 2024–2025 as financial services, scale-ups and the NHS Digital programme started shipping LLM-backed features in parallel. The Tech Nation 2025 report put UK open AI engineering roles at ~28,000 at end of 2025, against a senior (5y+) local supply of roughly 6,800 according to the same report — a 4:1 structural gap that the domestic graduate pipeline cannot close this decade. Public-university graduation of AI specialists (MSc + PhD combined) ran at ~1,400/year in 2024–2025; even absorbing every graduate does not close the gap.

Brexit has narrowed the EU-local pool for full UK employment. Sponsoring a non-UK EU engineer under the Skilled Worker route now takes 8–12 weeks and roughly £5K in legal / immigration fees per hire, and many UK scale-ups have quietly dropped the capability in the last three years. The practical alternatives are the B2B contractor model (no visa, outside IR35 for genuine contractors) and Employer-of-Record employment in the engineer's home country. Both work well — B2B is faster and cheaper, EOR offers better retention leverage. Our UK recruitment agency parent page walks through both in more detail.

IR35 is the concern most London CTOs raise first. Genuine B2B CEE contractors pass HMRC's outside-IR35 tests on every canonical axis: they control their own working pattern, they have their own registered business, they have multiple clients over a typical year, they supply their own equipment, and the contract sits between your UK entity and their overseas sole-proprietorship. For safety on inside-IR35 edge cases we default to EOR. We review the IR35 posture of every UK placement as part of onboarding; UK GDPR and EU AI Act reviews are done at the same checkpoint.

The role-specific angle the shortage literature misses: UK LLM engineers are not a subset of UK senior backend engineers with an extra course on prompting. The hires that actually move a product forward have spent 2+ years inside an LLM-backed product surface — they have opinions about prompt caching strategies, they have watched a cost spike trigger a weekend incident, they know why speculative decoding helped in one deployment and regressed quality in another. That kind of operator is scarce globally and disproportionately scarce inside the M25. The CEE senior pool we pull from is demographically weighted toward engineers who joined earlier AI companies (Neurons Lab, Wärtsilä AI, Allegro AI, Haut.AI) and therefore accumulated those production miles faster than a UK-local engineer who got exposure to LLMs mostly through their day job in 2024–2025.

Regional UK demand is also worth naming. London fintech and London-HQ enterprise SaaS dominate our pipeline, but Edinburgh (Scottish fintech + AI safety), Manchester (enterprise SaaS + health-tech), Bristol (aerospace + defence-adjacent), and Cambridge (research spin-outs) each open 2–4 UK LLM engineer roles per quarter that compete for the same thin senior pool. Outside those five cities, local hiring is effectively impossible above mid-level; UK scale-ups below the M25 rely almost entirely on remote EU hires. That is the precise gap CEE fills.