AI Harness Engineer

The Citation Group; Wilmslow; Hybrid; GBP 90,000 - 130,000; Full time

Job summary

The Citation Group is seeking an AI Harness Engineer in Wilmslow, with remote UK options. You will help build a system that orchestrates AI agents from brief to proven changes, with human review before anything ships.

You will shape how agents pick up work, handle failures, and improve the flow with observability, testing, and secure coding practices. This is a high-impact role central to our engineering strategy.

Qualifications

• Production experience building agent systems.
• Strong software engineering fundamentals.
• Experience with orchestration and tooling.

Responsibilities

• Design and build the harness to translate briefs into proven AI-driven changes.
• Ensure changes are verified beyond agent output and handle failures gracefully.
• Improve runs by measuring performance and implementing proven improvements.
• Help engineering teams adopt the harness and deploy agents in production.
• Write clean, tested, secure code with strong observability from day one.
• Uphold high engineering standards for AI-native systems.

Skills

Agent systems

Software engineering

Failure handling

Job description

AI Harness Engineer

Location: Wilmslow / Remote (UK)

About Citation Group

We are The Citation Group – a collective of businesses dedicated to supporting small and medium‑sized enterprises across a range of essential services.

We know that running a business means juggling a lot. Our mission is to ease these pressures by providing expertise, guidance and solutions that enable business leaders to focus on what they do best.

From HR and Health & Safety to Cybersecurity, E‑Learning and ISO compliance, we've got you covered.

Citation Group has achieved strong growth through a combination of organic expansion and strategic acquisitions, continually broadening our expertise, services and reach to create a one‑stop shop that supports businesses across the UK, Canada and Australia.

Role Purpose

We are changing how our engineering teams build software.

At the centre of that is the harness: the system that takes a well‑written brief and has AI agents plan, build, test and prove the work, with a person reviewing before anything ships.

You will build it.

You will shape how agents pick up work, how their output is checked, what happens when something goes wrong, and how every run makes the next one better.

This is one of the most consequential engineering roles in the group.

The right person has built agent systems for real, understands the software development lifecycle deeply, and holds agents to the same standard as any other engineer.

Key Responsibilities

Build the Harness

• Design and build the system that takes work from brief to proven pull request using AI agents.
• Make sure whatever builds a change is never the only thing checking it.
• Handle failure well: agents that stop when they should, and recover when they can.

Make It Better Every Run

• Measure how the harness performs, including success rates, rework and cost per change.
• Turn every failure into an improvement, and prove the improvement with evaluation.

Work With Teams

• Help engineering teams adopt the harness and get value from it quickly.
• Learn from teams already running agents in production elsewhere in the group.

Engineering Practice

• Write clean, tested and secure code, and take an active part in code review.
• Build with observability in mind from the start.

AI‑Native Engineering

• Use agents as a core part of how you build the system that runs agents.
• Hold agent output to the same standard as anything else we ship.

Required Experience and Skills

Technical

• Production experience building agent systems: orchestration, tool use and failure handling.
• Deep software engineering experience and understanding of