AI Solution Architect - FDT in Sheffield

Job Description

Level / Location: GCB4/5 typical— Various — co-located with the FDT Tech

Sit within the FDT Tech team and close to users (business teams, delivery pods, transformation squads) as the accountable architect for rapid shaping and delivery of AI-enabled, end-to-end solutions. Translate real user needs into target architectures spanning channels/UI, orchestration, AI services (ML/LLM), data stores, integrations, observability, and run operations—turning use cases into production-viable designs quickly, safely and repeatably. This role bridges product intent, engineering execution and governance, accelerating time-to-value while ensuring solutions are secure, supportable and aligned to enterprise standards and platforms.

Reporting & Stakeholders

Solid line into FDT Tech Lead / Engineering Lead. Strong day-to-day partnership with Product Owner(s), business SMEs and delivery pod leads on scope, prioritisation and trade-offs. Close collaboration with platform teams (e.g., AI platform / Data platform), AI/Data Engineers, and operations/support teams to ensure production readiness. Engage with tooling/platform owners to provide structured feedback and reusable patterns from field delivery.

Scope & Authority

Accountable for shaping and owning the solution architecture for AI-enabled thin-slice solutions close to users, including component selection, interfaces/contracts, non-functional requirements, controls-by-design, and lifecycle/run considerations. Owns local architecture decisions required to deliver outcomes at pace, within approved enterprise standards, reference architectures and governance frameworks. Authority includes defining build vs buy, selecting reference patterns (ML/GenAI/hybrid), producing SDDs, and escalating material risks, dependencies or control gaps.

Role Description & Core Accountabilities

This role exists because high-impact AI delivery needs fast, explicit architecture decisions: which AI pattern to use, how it plugs into workflows, how to manage safety/security/controls, and how to operate and improve the solution after go-live. The AI Solution Architect accelerates learning loops while maintaining architectural discipline, ensuring what’s designed can scale, be supported and be reused.

• End-to-end solution design (from problem to production) - Translate the use case into a target architecture: channels/UI, orchestration, AI services (ML/LLM), data stores, integrations, observability, and run operations. Define build vs buy, component selection, and non-functional requirements (availability, latency, cost, resilience).
• AI pattern selection and fit-for-purpose design - Choose the right AI approach and reference pattern: predictive ML (batch/real-time scoring), GenAI (RAG, summarisation, classification with LLMs, agentic workflows/tool use), and hybrid designs (rules + ML/LLM). Make the “why this approach” explicit: expected uplift, risk profile, operational complexity.
• Integration into the value stream - Design how AI outputs become decisions/actions in the workflow (human-in-the-loop vs straight-through processing). Define interfaces and contracts with upstream/downstream systems (APIs, events, data contracts), including error handling and fallbacks.
• Security, privacy, and controls-by-design - Ensure authentication/authorisation, data minimisation, encryption, logging, and secrets management are built in. Prevent data leakage and injection risks; ensure controlled tool access and safe prompt/context handling. Align solution to required governance (model approval, validation evidence, audit trail); coordinate with risk/compliance partners.
• MLOps/LLMOps and lifecycle management - Define deployment strategy, model/version management, prompt/versioning, and rollback approaches. Specify monitoring: model quality, drift, retrieval quality, safety metrics, latency, cost, and business KPIs. Define retraining/re-indexing cadence and ownership (RACI) post go-live.
• Architecture governance and technical leadership - Own architecture decisions (ADRs), ensure alignment to enterprise standards, and manage technical debt. Provide cross-squad reuse: reference architectures, templates, common components (e.g., evaluation harness, guardrails service).
• Delivery enablement - Break architecture into deliverable increments and help the Product Owner/Scrum team sequence work. Unblock engineers with clear designs, diagrams, interface specs, and decision logs.

Calibration Rationale

This is a hands-on, user-adjacent architecture role optimised for speed-to-value and safe, operable AI-enabled delivery. Core capabilities around architectural judgement, customer centricity, integration design, controls-by-design and pragmatic decision-making are Essential. Broader strategic leadership and talent stewardship are Desirable, as the role contributes to scaling patterns and mentoring but does not typically own enterprise-wide strategy.

Experience — Required

• Proven experience designing end-to-end solution architectures in enterprise environments, covering channels/UI, APIs/integration, data stores, observability and run operations.
• Strong ability to rapidly shape options with squads and users, then produce production-viable target architectures and incremental delivery plans.
• Experience selecting fit-for-purpose AI patterns (predictive ML, GenAI/RAG, hybrid) and articulating trade-offs (uplift vs risk vs operational complexity).
• Experience defining non-functional requirements (availability, latency, resilience, cost) and translating them into concrete design decisions and SLOs.
• Experience defining interfaces and contracts (APIs, events, data contracts), including error handling, fallbacks and degraded-mode behaviour.
• Working knowledge of security, privacy and controls-by-design practices (authentication/authorisation, encryption, secrets, logging, data minimisation) and how to embed them in architecture.
• Comfortable operating in ambiguous environments with shifting requirements, while maintaining delivery discipline and transparency.

Experience — Preferred

• Experience delivering within regulated/controlled environments (risk, audit, security, data handling) and navigating governance pragmatically (approval evidence, audit trails).
• Experience with GenAI architectures including RAG/agentic workflows, guardrails, grounding/citation patterns, and evaluation approaches.
• Experience with MLOps/LLMOps, including versioning, deployment strategies (canary/blue-green), monitoring, rollback and lifecycle ownership models.
• Experience contributing reusable reference architectures, templates and patterns across multiple squads/value streams.
• Experience working in Value Streams / product-led delivery, transformation pods, or embedded/forward-deployed models.

Behavioural Skills

• User-obsessed architect; stays close to users and delivery teams; designs for real workflows, not just diagrams.
• Pragmatic architecture discipline; moves quickly without skipping fundamentals (security, resilience, operability, controls).
• End-to-end ownership mindset; takes accountability from discovery through build, release and run; doesn’t throw designs “over the wall”.
• Evidence-led trade-offs; makes “why this design” explicit; uses measurable NFRs, risk assessments and evaluation plans to justify choices.
• Clear communicator under pressure; explains options, trade-offs and risks in plain language; keeps stakeholders aligned and decisions unblocked.
• Collaborative and inclusive; works well across roles and backgrounds; welcomes different perspectives to reach stronger designs.
• Proactive risk management; identifies integration, safety, dependency and control risks early; escalates with options, not surprises.

Derailers / Red Flags

• Creates target-state architectures that are not deliverable in increments, causing slow starts and rework.
• Over-optimises for elegance or novelty (especially in GenAI) without proving business value or operational feasibility.
• Ignores security, privacy, controls or operational needs in the name of speed; designs fragile solutions into production pathways.
• Creates bespoke point solutions where reusable platforms/patterns exist.
• Fails to define clear interfaces, fallbacks and degraded-mode behaviour, leading to brittle integrations.
• Treats monitoring, runbooks and lifecycle ownership as an afterthought, increasing post-release risk.

Year 1 Outcomes (Scale & Embed)

• Multiple AI-enabled solutions designed and delivered with clear architecture packs, SDDs and measurable NFRs met.
• Reusable reference architectures/templates established for common AI patterns (predictive scoring, RAG, hybrid decisioning), reducing time-to-start for squads.
• Consistent controls-by-design embedded into AI solution designs (security, privacy, audit trail), reducing governance rework and late-stage blockers.
• MLOps/LLMOps lifecycle patterns embedded (versioning, deployment, monitoring, rollback, RACI), improving reliability and speed of iteration post go-live.
• Improved delivery flow through clearer increments, interfaces and dependency management; fewer late surprises during integration and release.
• Recognised as a dependable embedded architecture partner who unblocks delivery and raises overall engineering maturity.