SVP – Engineering — Data And Evaluation Platform

The mandate

The data and evaluation platform serves research teams and external customers through multiple model variants. As demand and use cases have grown, the platform now requires clearer architecture principles, simplified delivery commitments and stronger technical leadership. The SVP – Engineering will establish coherent platform architecture, reset roadmap priorities and build an engineering organisation capable of predictable delivery at scale.

The remit spans approximately £1.05 billion in AI product and services revenue and about 650 employees and material partners across the UK, London and the wider operating region. Engineering covers data pipelines, evaluation systems, platform services, developer tooling, reliability and customer integrations. The organisation possesses substantial expertise; its constraint is how priorities, dependencies and architectural standards translate into work.

The mandate calls for a deliberate change of pace. Near-term customer and safety obligations must remain stable, while low-value work is stopped and foundational constraints receive concentrated capacity. Roadmap predictability cannot be achieved by lowering ambition or hiding uncertainty. It requires explicit choices, smaller increments and reliable measures of throughput, quality and consumption.

Why this seat is open

This is a newly created urgent post because model-cost escalation and excess commitments need one engineering executive rather than distributed arbitration. Interim governance maintains immediate delivery, but permanent leadership is required within six to eight weeks. The onsite London role supports international relocation and reports to the Group Chief Executive or nominated executive committee sponsor.

What you will own

You will establish a reconciled portfolio of commitments, dependencies, capacity and technical debt. Work should be classified by customer obligation, platform value, risk reduction and experimentation. Product and commercial leaders must see the capacity and architecture consequence of requests before promising them. The executive team needs explicit trade-offs where demand exceeds supply.

Architecture coherence will be managed through durable boundaries, decision records and empowered technical leaders. You will identify where variants reflect real product needs and where they are symptoms of weak platform interfaces. Model, compute and evaluation consumption should be observable at service and workload level, enabling engineering choices based on total economics.

Delivery discipline must move from activity to flow. Limit concurrent work, expose queue and rework, improve test and release automation and use incidents as architecture evidence. Reliability and security are engineering outcomes, not downstream control functions. A roadmap date should express known dependencies and confidence, not an aspirational commitment detached from teams.

The organisation requires stronger leadership below the SVP. Assess executives, principal engineers and site leaders; clarify who owns architecture, delivery and people outcomes; and build successors for critical seats. Scarce engineers should work on the most consequential constraints rather than repeatedly rescuing local launches. Partner capacity must meet the same quality and evidence standards.

The first 12 months

During the first 90 days, reconcile commitments with available capacity and trace the main sources of delay, rework and model cost. Review important incidents, architecture decisions and customer promises. Assess leadership, protect critical delivery and bring the board a reset roadmap, investment gates and the decisions required to stop or sequence work.

From months four to nine, implement portfolio limits, strengthen architecture ownership and address the largest technical constraint. Improve engineering measures and release practices, fill pivotal roles and demonstrate a material delivery path with better predictability, quality and model economics. Exceptions should have owners, expiry and quantified cost.

At year end, roadmap reliability and technical quality should be improving across multiple teams, supported by a stronger leadership bench. The next annual plan needs one view of customer demand, engineering capacity, architecture investment and workforce. Present a three-year technology case with downside moves if model cost or delivery performance differs from assumptions.

What the board will measure

First-year engineering commitments should remain within 10% of approved outcomes and investment, with variance forecast ahead of failure. Three consecutive quarterly views must align revenue, cash, customer delivery, model capacity and people. A named cause of cost escalation or roadmap instability should improve measurably from an agreed baseline.

Critical quality, reliability and architecture issues must close by authorised dates and survive subsequent testing. At least 90% of pivotal engineering talent should remain, and 70% of direct reports should have ready-now successors. Severe technical escalations cannot be left undecided for more than 30 days or concealed from the appropriate forum.

The person

You are an SVP Engineering, VP R&D or Engineering Centre Head with 22–28 years in AI, enterprise software, data infrastructure, cloud, analytics or applied research. You have owned at least £750 million in P&L, budget, book or accountable engineering portfolio and led at least 650 people.

Your evidence includes a commitment reset that improved delivery without weakening a strategic platform. You can show how architecture, flow, model economics and leadership changed, with numerical effects sustained beyond a release. The board will explore what you stopped and how you handled disappointed stakeholders. References must distinguish your choices from broader product or market developments.

Compensation and terms

The anticipated package is £210,000–280,000 base plus annual incentive, calibrated to final scope and current mix. Any longer-term participation uses standard vesting and performance terms. A notice period up to six months is acceptable. The SVP receives recurring exposure to the group board and relevant risk and people committees.

Confidentiality

The company, architecture, customer commitments and interim leadership remain confidential. Identifying details will be shared with qualified candidates under an undertaking; public circumstances are rounded and blended.