Build production AI at sovereign scale

The job titles read similarly across the market, but the underlying briefs split into three clusters:

1. Sovereign LLM teams — TII, G42 Inception, HUMAIN. Pre-training, post-training, evaluation. Genuine research adjacency, with the expectation that you can read and contribute to recent literature.
2. Applied AI inside banks and energy majors — Aramco Digital, Emirates NBD, ADNOC AIQ. Fine-tuning, RAG, agentic workflows for internal use cases. Less novel, more delivery pressure.
3. Product engineering inside G42 portfolio operating companies — Presight, Core42. Customer-facing AI, MLOps maturity, multi-tenant deployment. Closer to a Silicon Valley product team in profile.

A candidate strong in cluster 1 is often a poor fit for cluster 3, and vice versa. The mistake on both sides — clients and candidates — is treating "Senior ML Engineer" as a single role. We screen briefs into these three categories before opening a search.

Most teams are still small. Senior ICs typically report directly to a Head of AI or VP of Engineering, ship to production within their first quarter, and own a stream of work end-to-end. Engineering scaffolding — CI/CD, observability, evaluation harnesses — is usually half-built. If you have only worked at organisations where this was already in place, expect the first six months to feel less polished than your last role.

The flipside is autonomy and scope. A Senior ML Engineer at a UK bank might own one component inside a much larger platform team. The same engineer at a GCC bank typically owns an entire system — model, serving, evaluation, monitoring — and is in the conversation when the next system gets scoped.

"The supply–demand gap at 8–15 years of experience is the widest we have seen in any market."

Three reasons:

1. The depth bar is higher than the title implies. Clients say "Senior" and mean what most US tech firms would title Staff. Genuine production ML experience, not API-call experience.
2. The candidate pool that fits the brief and is genuinely open to a GCC move is much smaller than LinkedIn search suggests. Most engineers with the right technical profile have never considered the region.
3. Most strong candidates are talking to two or three other employers in the region by the time you reach them. The market is small and connected — recruiters at TII, G42, and HUMAIN are working from overlapping shortlists.

Total compensation is rarely the deciding factor at offer stage. Schooling, spouse career, and visa flexibility decide more offers than salary at this level. Clients who lead with the right relocation package — including specific school placements, spouse work authorisation, and a clear ninety-day arrival plan — typically close 30–40% faster than those who lead with the highest base.

For candidates already in the region, the calculus is different. Counter-offers from the current employer are common and frequently substantial. We tell candidates to model the next 24 months, not the next 12 — most of the value of a move is in trajectory, not the first-year delta.

The product side. Most candidates assume the GCC opportunity is technical depth. The bigger differentiator is that you are typically two reporting layers from the CEO of a meaningful business, with a real budget and a real timeline. Decisions move faster than equivalent roles at FAANG-scale organisations, and individual contributors have more visibility into strategy than they would at home.