Overview of Role
This is a senior engineering role on an EMEA Change Team delivering a next-generation policy administration system for a leading global insurance organisation. This role is responsible for designing, building, and governing AI agent infrastructure across the programme. It combines agent orchestration, prompt contract design, RAG pipelines, quality evaluation, and guardrails to ensure reliable production use of LLM-based systems. It also supports delivery engineers by making agent tooling usable, measurable, and safe at scale.
Role & Responsibilities
Design and maintain AI agent workflows, sub-agent orchestration logic, and tool integrations for feature delivery across the programme.
Author and version prompt contract templates and schemas to ensure consistency and quality across delivery work-streams.
Build and maintain knowledge base ingestion pipelines and RAG implementations that ground agents in accurate, programme-specific domain knowledge.
Own cost telemetry, model tier-routing rules, evaluation harnesses, guardrails, and safety mechanisms for production agent pipelines.
Co-author the AI agent contract section of technical design documents, validate contract structure before build begins, and support delivery engineers with tooling integration and contract interpretation.
Skills, Qualifications & Experience
Strong hands-on AI/ML experience with a focus on LLM applications in production environments.
Proven experience designing multi-step agentic workflows with tool use and sub-agent orchestration.
Strong prompt engineering capability, including structured prompt contracts for controlled output at scale.
Experience with automated evaluation frameworks, RAG pipelines, knowledge base ingestion, and production guardrails.
Proficiency in Python, with familiarity with Claude API or equivalent, MCP, and cost control practices.
Experience
5+ years of hands-on AI/ML experience, with production LLM delivery rather than prototype work.
Experience with agent frameworks such as Claude Agent SDK, LangGraph, or equivalent.
Experience building automated evaluation harnesses for precision, recall, hallucination rate, and regression detection.
Experience with vector databases such as pgvector or Qdrant, and with AI tooling integration in developer workflows.
Education & Certifications
Bachelor's or Master's degree in Computer Science, AI, or a related field.
Relevant AI/ML certifications are beneficial but not required.
#J-18808-Ljbffr