AI Product Architect

Location: UK/Hybrid

Contract type: 6 months (interim with strong possibility of extension)

Role Purpose

The AI Product Architect is responsible for translating complex business challenges into scalable, production-grade AI solutions. Working at the intersection of business strategy and engineering delivery, this role leads the end-to-end design of AI systems from integration with enterprise platforms and legacy data estates through to the orchestration of autonomous agentic workflows and ensuring every solution is governed, explainable, and aligned with enterprise risk and compliance standards. Strategizes and designs enterprise-level AI integrations, manages AI governance and risk, and acts as the primary liaison between technical teams and business stakeholders.

Key responsibilities:

Legacy & Enterprise integration

•    Design integration patterns that surface data from legacy systems into modern AI pipelines, including ETL/ELT patterns, API wrappers, and event-driven architectures.
•    Integrate AI products into enterprise ecosystems including Workday, ServiceNow, SAP, Salesforce, AWS, Azure, and Snowflake.
•    Define data contracts and transformation standards that ensure consistency and quality across source systems.

Workflow and process transformation

•    Translate ambiguous business problems into discrete, autonomous decision chains with clearly defined inputs, outputs, and failure modes.
•    Identify and design human-in-the-loop approval gates, escalation paths, and override mechanisms within agentic workflows.
•    Map current-state processes and produce target-state AI-enabled operating models in collaboration with business stakeholders.

Agentic framework design and delivery

•    Architect and implement multi-agent systems using orchestration frameworks including LangGraph and CrewAI.
•    Define agent roles, memory strategies, tool-use patterns, and inter-agent communication protocols.
•    Evaluate and select appropriate agentic architectures (single-agent, hierarchical, collaborative) for each use case.

LLM & RAG Architecture

•    Design end-to-end Retrieval-Augmented Generation pipelines including document ingestion, chunking strategy, embedding model selection, and vector store configuration.
•    Apply hybrid search (dense + sparse), re-ranking, and query expansion techniques to optimise retrieval quality.
•    Model knowledge graph structures to enhance contextual reasoning and relationship traversal across enterprise data.
•    Understand and account for LLM behaviour including context window management, temperature tuning, hallucination mitigation, and prompt engineering.

AI Governance and Risk

•    Design end-to-end Retrieval-Augmented Generation pipelines including document ingestion, chunking strategy, embedding model selection, and vector store configuration.
•    Apply hybrid search (dense + sparse), re-ranking, and query expansion techniques to optimise retrieval quality.
•    Model knowledge graph structures to enhance contextual reasoning and relationship traversal across enterprise data.
•    Understand and account for LLM behaviour including context window management, temperature tuning, hallucination mitigation, and prompt engineering.

Stakeholder Management

•    Act as the primary bridge between business, product, engineering, data science, security, and compliance teams.
•    Communicate complex technical architectures to non-technical audiences through clear documentation, diagrams, and presentations.
•    Lead architecture reviews, design workshops, and proof-of-concept evaluations with cross-functional teams.
•    Provide expert guidance and challenge to senior stakeholders on AI feasibility, risk, and value delivery.

Skills and Experience Required

• Legacy Integration: Proven experience designing integration patterns to ingest and transform data from legacy systems into AI-ready formats.
• Platform Integration: Hands-on experience integrating AI solutions into two or more of: Workday, ServiceNow, SAP, Salesforce, AWS, Azure, Snowflake.
• Workflow Transformation: Demonstrated ability to decompose business processes into autonomous decision chains and design appropriate human oversight mechanisms.
• Agentic Frameworks: Practical experience building production systems with LangGraph and/or CrewAI, including multi-agent coordination and tool orchestration.
• LLM & RAG: Deep knowledge of embeddings, chunking strategies, hybrid search, vector databases, knowledge graphs, and LLM behaviour.
• AI Governance: Experience implementing responsible AI controls, explain-ability mechanisms, and compliance frameworks in production environments.
• Stakeholder Management: Ability to operate across technical and business stakeholders, translating between strategic objectives and engineering delivery.

The strongest candidate will demonstrate the following

•    A portfolio of AI systems delivered into production with clear descriptions of architecture decisions and trade-offs made.
•    The ability to walk through a live agentic workflow they designed, explaining how they determined agent boundaries, tool selection, and approval gates.
•    Experience of integrating AI into at least one named enterprise platform (e.g. ServiceNow, Salesforce) and the ability to describe the specific integration challenges encountered.
•    Confidence discussing AI governance at a practical level, not just policy awareness but actual controls implemented, monitored, and audited.
•    Strong written communication: architecture decision records, LLDs, and stakeholder presentations that demonstrate clarity of thought.