Agentic Engineer (six-month contract, outside IR35)

Job title: Agentic Engineer (six-month contract, outside IR35)
Type: Day-rate contractor, outside IR35
Location: London, client site 2 to 3 days a week
Start date: Monday 28th September, must be available to start then

We're looking for an Agentic Engineer to deliver a six-month engagement for hedgehog lab, embedded with a large UK content and media client, building the AI-powered workflows that put agentic engineering into practice for them. What you build, and how well you document it, becomes the template for the client's permanent hire once this engagement ends.

You'll turn business and operational problems into agentic systems that hold up in production. Reliable, observable, useful.

You'll also use AI agents to produce a meaningful share of your own code, tests, and documentation, directing and checking AI-generated work rather than writing everything by hand. Same quality bar as anything hand-written.

About hedgehog lab

UK digital product consultancy and certified B Corp, with offices in Edinburgh, Newcastle, Leeds, Manchester and London.

What you'll deliver

• Map workflows with users and cross-disciplinary teams, and define clear outcomes for agentic solutions

• Design and build agentic systems: orchestration, tool use, retrieval and memory, human approval, failure recovery

• Integrate AI capabilities with the client's platforms through APIs, events, and data services

• Build reusable components and patterns for the client's team to reuse

• Build automated evaluations and tests covering task success, quality, accuracy, and safety

• Set up observability and tracing, and improve agent behaviour, latency, and cost

• Apply security, privacy, and content-rights controls, with human review built in

• Document decisions clearly enough for the client's permanent hire to pick up from

Who this is for

Comfortable owning ambiguity, with no client project manager directing your day-to-day. You'll set your own approach and be judged on outcomes, with a fixed six-month end date and handover in mind from day one. Able to start Monday 28 September.

Experience you'll need to be considered

• Strong software engineering, with clear, maintainable, well-tested code

• Hands-on experience building applications with LLMs and agentic patterns, beyond prompt-only prototypes

• Tool and function calling, structured outputs, RAG, context management, human-in-the-loop design

• Experience integrating APIs, data services, and third-party platforms into reliable end-to-end products

• Experience designing automated evaluations and tests for non-deterministic AI systems, including quality, safety, and regression testing

• Familiarity with observability, tracing, and production monitoring for AI-powered applications

• Experience deploying and operating cloud-based services using modern CI/CD and software delivery practices

• Good understanding of application security, privacy, responsible AI, and safe handling of content and data

• Genuine day-to-day experience using agentic AI tools for your own engineering output, with a track record of directing and validating AI-generated code, tests, or documentation

Nice to have

• Background in media, publishing, or content-driven businesses

• Familiarity with MCP integrations, agentic systems, or LLM-powered applications