Our client is a fast-growing European fintech company in the business spend management space — corporate cards and related financial products — serving SME and mid-sized business customers across the EU and UK, in a regulated environment with GDPR compliance obligations. Engineering is organized into cross-functional squads, with platform/enabling teams providing shared tooling horizontally. The client already runs an AI-native engineering practice: Claude Code and GitHub Copilot are used daily, supported by a growing library of shared, versioned Skills embedded in key repositories, enabling an end-to-end flow from ticket to implementation, testing and PR in several codebases. AI tool/connector rollout follows an approved-list and pilot process. The client is AWS-first overall; for data platform and analytics workloads it also runs GCP, with BigQuery as the primary data warehouse. Day-to-day coordination is Slack-first, with Linear for ticket tracking, GitHub for code/PR review, and Notion as the knowledge base.

The client is also building out its AI platform and agentic capabilities: it recently launched an MCP (Model Context Protocol) surface in closed beta, giving external AI assistants a structured way to connect and perform real workflows behind guardrails, and treats agent reliability as a system property (tool contracts, approval/consent gates for write actions, evaluation harnesses).

About This Role

We're looking for a Senior, Python-first Applied AI Engineer to design, build and ship customer-facing AI features end-to-end — not prototypes, production.

Key Responsibilities

Design, build and ship customer-facing AI features end-to-end, including RAG system design (chunking, embedding selection, retrieval, re-ranking) and agentic workflows.
Ship external customer-facing LLM-based features into production, with safe production deployment practices.
Do hands-on evaluation-pipeline and observability work: drift detection, quality monitoring.
Reason independently about retrieval architecture using strong data fluency.
Apply real engineering rigor to the Context Development Lifecycle (CDLC): generate, evaluate, distribute and observe the context that powers AI agents.
Multiply the output and quality of the squad you join, and share patterns/practice beyond your immediate team so adoption compounds across the organisation.
Embed directly into a client squad as a hands-on Individual Contributor (not a coaching/managerial role).

Must-Have Skills

Python (Python-first).
RAG system design: chunking, embedding selection, retrieval, re-ranking.
Agentic workflows; safe production deployment.
Proven experience shipping external customer-facing LLM-based features to production (not prototypes).
Hands-on evaluation-pipeline and observability work: drift detection, quality monitoring.
Data fluency to reason about retrieval architecture independently.
Agent & harness engineering fluency — designing guardrails, context and verification layers for safe, reliable AI-assisted delivery; concrete evidence of having built/operated this kind of tooling, not just used it.
Comfortable with the client's current daily tools: Claude Code and GitHub Copilot.

Nice-to-Have Skills

Specific tooling such as AWS Bedrock, OpenAI APIs, or Langfuse (or comparable eval platforms). These are not named explicitly in the RFP — listed here only as illustrative, commonly-used tooling in this space, not a confirmed client requirement.

Applied AI Engineer

Job Description