AI Engineer / AI Architect

About the Role:
We're looking for a forward-deployed AI engineer to embed with the existing analyst team and transform how they work with LLMs - day to day, in practice. This is as much an elicitation and communication role as an engineering one: you'll sit with the analysts, learn their workflows deeply, and turn that understanding into working AI-powered tooling.
The immediate priority is connecting our models (Claude and Gemini) to external data sources via MCP, leveraging existing servers where available, writing new ones where needed, and building practical agentic tools and research "skills" the team can use immediately. This will feed into a broader LangGraph-based platform already under development, but extracting real value for our analysts now comes first.

What You'll Do:
- Embed with analysts to understand their investigative workflows and translate them into working tooling.
- Connect LLMs to external/third-party databases and internal resources via MCP.
- Build practical agentic tools and research skills across Claude and Gemini.
- Advise on best use of our current models and agent architecture.
- Review and provide recommendations on existing internal development efforts.
- Help define how LLMs integrate into the team's daily work — before automating for external-facing systems.

What We're Looking For:
- Forward-deployed / solutions engineering experience, a track record of embedding with non-technical specialists and shipping tools for them. This is the most important qualification.
- Hands-on experience with both Claude and Gemini: tool/function calling, agentic patterns, MCP.
- Comfort treating prompt engineering and agent/skill design as core work, not a side task.
- Strong security awareness.
- Excellent English communication skills (written and verbal).
- Python proficiency (FastAPI, async patterns).

Nice to have:
- Familiarity with LangGraph is a plus 
- React/TypeScript familiarity.

About the Client
This is a specialist intelligence and investigations firm whose analysts conduct deep-dive research into complex entities: individuals, organisations, and networks, etc. for clients operating in high-stakes environments: due diligence, compliance, litigation support, and financial crime. The work is detail-intensive and source-critical: every finding must be traceable, verifiable, and defensible. They are now building AI infrastructure to support and accelerate that process - not to replace human judgement, but to give their analysts better tools.

Stack: Python 3.10+, FastAPI, LangGraph, Pydantic | React, TypeScript, Vite, Zustand.