AI Solutions Lead — Agentic AI & RAG

The mandate

You will lead the AI Solutions practice on programmes where the model is the easy part. The hard work — retrieval quality, guardrails, audit trails, human-in-the-loop placement, latency under SLA, fallback design — is where you spend your time. Engagements range from banking compliance copilots to government regulatory assistance and life-sciences document automation. Pilot-to-production is the work; demos are not the outcome.

Responsibilities

• Architect RAG pipelines with retrieval evaluation, reranking, and guardrails as designed boundaries (not bolt-ons)
• Lead agentic AI builds with tool-use orchestration, planner-executor patterns, and human checkpoints at consequence
• Establish prompt + model versioning, evaluation gates, and rollback paths for regulated change management
• Own the AI-native UX patterns alongside design — conversational, explainable, deferrable
• Coach mid-level AI engineers on the eighty percent that ships AI to production

Requirements

• 8+ years engineering with at least 2 years leading production LLM or RAG systems for regulated clients
• Deep familiarity with LangChain or LlamaIndex, vector stores, and reranking models
• Hands-on experience with guardrail frameworks, eval harnesses, and immutable audit logging
• Track record translating compliance constraints into AI architecture, not vice versa
• Comfort sizing AI engagements honestly and saying no to use cases that should not be automated