Hire ML platform engineers in the UK — 8-day shortlist, IR35-safe.

Worked example, senior ML platform engineer into a PRA-regulated London insurer (placed 2026-Q1). Target role: staff-level ML platform engineer owning Kubeflow, Feast feature store, and SageMaker-to-self-hosted migration. UK-local equivalent would have been £162K base + 18% loaded overhead (NI, pension, on-call, equipment) = £191K/yr. Actual placement: CEE senior platform engineer at €98K/yr on B2B contract (≈£84K). Recruo fee: 15% of £84K base = £12.6K one-time. Classical UK agency alternative fee: 22% of £162K UK-local = £35.6K. Year-1 total delta to the client: (£191K + £35.6K) − (£84K + £12.6K) = £130K. From year 2 onwards the gap compresses to roughly £107K/yr once the one-time fee falls off.

ML platform engineering is the pointed end of the UK AI talent shortage. Two shortages compound: infrastructure-specialist engineers were already scarce in the UK market before 2024, and the sub-population who have run production ML infra at scale — Kubernetes-native training orchestration, feature stores, online/offline parity, GPU scheduling, model serving at P99 latency, on-call maturity — is a small fraction of that. LinkedIn Talent Insights puts open London ML-platform roles at roughly 2,400 at end of Q1 2026 against a senior local supply of around 480, a 5:1 structural gap that is materially tighter than the 4:1 seen on LLM-engineer roles. Domestic pipeline does not close it — UK universities produce well under a hundred graduates per year who have touched anything resembling production MLOps, and most are immediately hoovered up by DeepMind, the ex-OpenAI London cluster, and the larger US hyperscalers.

UK demand sits disproportionately in regulated verticals. Revolut, Monzo, and Starling are all running internal ML platform teams; JPMorgan London, Schroders, and the larger insurers are hiring platform engineers into risk, surveillance, and claims-ML teams; Darktrace and Wise need platform engineers to keep training pipelines compliant with UK GDPR data-residency; the BBC and Sky have in-house recommendation platforms; GDS and the NHS-Digital programme now ship ML-backed services in production. Each of these seats requires an engineer who understands UK operational-resilience rules — FCA SYSC 15A, PRA SS1/21, and in financial services the Critical Third Party designation that is live since 2025. CEE engineers with prior exposure to EU DORA (which UK regulators have substantially aligned to) transfer cleanly; our screening confirms it before shortlist.

Brexit narrowed the EU-local full-hire pool, and IR35 changes from April 2021 reshaped how platform engineering specifically gets engaged. IR35 worries London CTOs more than any other compliance topic because platform engineers often embed tightly with in-house teams, which superficially looks like inside-IR35 risk. In practice our genuine B2B CEE contractors pass HMRC's outside-IR35 canonical tests on every axis — substitution right, working-pattern control, own registered business, multiple clients across a year, own equipment — and the contract runs from your UK entity to their overseas sole-proprietorship, not your payroll. For inside-IR35 edge cases, or FCA-regulated clients who want a single-payroll record for SYSC evidencing, we default to Employer-of-Record in the engineer's home country. Our UK recruitment agency parent walks through both models in more depth.