Senior AI Engineer

Key Responsibilities

• Agentic AI: Design, build and operate the agent-based and multi-agent systems behind our reasoning workflows. Build and run agentic systems in production: tool use, controller–executor architectures, orchestration, state and memory, and failure recovery. Improve how our agents create and enrich attributes on demand (e.g. turning a user’s prompt into a structured attribute) and keep them consistent with our ontology. Contribute to the tooling we use to define and extend our ontology with clients.The expectation is that you own the difference between an agent that demos well and one that holds up under real usage.
• Evaluation: Bring rigour to how we judge the quality of our AI systems. Choose the right datasets, metrics and methods (offline and online, human and LLM-as-judge) and know the limits of each. Trace a bad answer back to its cause. Make quality measurable across our AI systems. For every client-facing workflow, build an evaluation set with the client’s domain experts before it ships, and gate releases on it. Model and prompt changes should ship on evidence rather than intuition.
• Search & Retrieval: Own and advance our search and retrieval layer. Drive indexing strategy, hybrid lexical/semantic retrieval, ranking and relevance quality. Extend the layer to index the client’s own documents and data. Apply access controls that keep permissioned content separate. The retrieval layer is the foundation the agents stand on: relevance and permissions both have to hold.
• Client Use Cases: Turn the client’s priority use cases into working agentic workflows. Join discovery sessions with the Senior Solutions Architect and the client. Prototype the client’s priority use cases on our core product within days. Work with the Senior Solutions Architect to take them from prototype to production. Everything built for the client is built directly on our core product.
• Software Foundations for Data Science: Be the software engineering counterpart to the data scientists working on scoring. Turn research code into reliable, observable, tested services and pipelines. Provide the libraries and interfaces that let data scientists move from experiment to production without rebuilding the plumbing each time. Data scientists should be able to take a model to production without becoming software engineers.
• Cost & Reliability: Keep our AI workflows affordable and dependable under real client usage. Monitor and reduce the LLM and infrastructure cost of each workflow. Keep agents reliable through fallbacks, retries and observability. A workflow is only finished when its cost and reliability are known and under control.
• Cross-functional Collaboration: Integrate AI capabilities cleanly into the platform and raise the engineering bar. Work closely with Data, Software, Product and Solutions, as well as the client. Contribute to code reviews and share practices. Raise the engineering bar across the AI team through collaboration and example. You will not manage a team; you will shape how the team works through the systems you build, the reviews you give and the standards you set.

What Success Looks Like

Success in this role will be visible in the quality and reliability of the AI workflows the client uses.

• Within the first 30 days, the Senior AI Engineer will have:
  • been onboarded on our platform and agent stack;
  • agreed the client priorities with the team; and
  • drafted an evaluation approach.
• Within the first 60 days, the Senior AI Engineer will have:
  • proven the complex AI components on the client’s data and validated them with the client; and
  • put an evaluation set and quality baseline in place.
• Within the first 6 months, the Senior AI Engineer will have:
  • taken the agreed workflows live in production with the client and its first users;
  • ensured every release passes the evaluation set; and
  • tracked quality, latency and cost per run.

What We Are Looking For

This role requires a combination of production AI engineering, rigour in evaluation and pragmatism with real users.

We are looking for someone who:

• Has a Master’s degree or PhD in computer science, mathematics, physics or data science.
• Has 7+ years in software engineering, including 2+ years building LLM-based systems, with hands-on ownership of production-grade LLM or ML systems.
• Has practical experience building multi-agent or tool-using architectures (prompt engineering, RAG, controller–executor patterns) and operating them in production.
• Can point to at least one agentic system they shipped that real users relied on and explain how they measured it.
• Has hands-on Elasticsearch (or equivalent) experience: index design, query tuning, hybrid and semantic retrieval, relevance evaluation.
• Is expert in Python, with a deep understanding of modular design, clean code and testable architectures.
• Has solid backend skills: API design, async processing and containerised deployment.
• Is familiar with LangGraph or comparable agentic frameworks (LangChain, LlamaIndex, OpenAI Agents SDK, Claude Agent SDK).
• Has built evaluation pipelines for LLM systems (golden datasets, LLM-as-judge calibrated against human review, regression tests in CI).
• Takes a problem end to end, including the unglamorous parts.
• Communicates clearly and concisely; is comfortable working with researchers and engineers alike and can explain AI trade-offs to non-technical client stakeholders.
• Solves problems pragmatically, with a bias for action in a fast-moving environment.
• Makes the people around them better through example, without needing authority to do it.
• Experience with document and PDF processing at scale (parsing, extraction, structuring), structured reasoning combined with LLMs, productionising scoring or ranking models, energy, sustainability or financial data, or working directly with business users to turn loosely defined needs into working AI features would be valuable. However, a track record of shipping AI systems that real users rely on matters more than any background.

Our Tech Stack

What you will work with day to day:

• Python
• LangGraph, LangChain / LangSmith
• ElasticSearch
• Azure Cloud
• Claude / Anthropic models and managed agent.

The wider platform your work plugs into:

• TypeScript, Angular / NestJS, Cypress