AI Harness Engineer

Morson Edge (Technology); Manchester; Hybride; GBP 75 000 - 110 000; Plein temps

Résumé du poste

Agentic Infrastructure in Manchester is seeking an AI Harness Engineer to take autonomous AI agent infrastructure from prototype to production. You will build secure harnesses around AI models, runtimes, guardrails and tooling to run at enterprise scale.

Key focus areas include sandboxed environments for agent code execution, prompt-injection protection, tool orchestration between LLMs and external systems, verification loops and scalable, secure runtime architectures.

Qualifications

• Strong software engineering or AI infrastructure background.
• Experience with Docker / Kubernetes and WASM or micro-VMs.
• Solid API integration and systems engineering experience.
• Good understanding of LLMs / generative AI / agentic systems.
• Security-first mindset around prompt injection and runtime security.

Responsabilités

• Building secure, sandboxed environments for AI agents to execute code
• Designing guardrails around agent behaviour, including input/output filtering and prompt injection protection
• Building tool orchestration and validation layers between LLMs and external systems
• Developing verification and self-correction loops around runtime and tool failures
• Designing human-in-the-loop checkpoints for higher-risk autonomous actions
• Making sure agent infrastructure is scalable, resilient and secure

Connaissances

Docker

Kubernetes

WASM

Micro-VMs

APIs

LLMs

Generative AI

Agentic systems

Security

Reliability

Description du poste

AI Harness Engineer | Agentic Infrastructure | Manchester hybrid

I’m working with a team building the infrastructure behind autonomous AI agents and they’re looking for an AI Harness Engineer to help take this from prototype into production.

You’ll be building the harness around AI models, the infrastructure, guardrails, tooling and runtime environments that make autonomous agents secure, reliable and scalable in an enterprise environment.

What you’ll be working on:

• Building secure, sandboxed environments for AI agents to execute code


• Designing guardrails around agent behaviour, including input/output filtering and prompt injection protection


• Building tool orchestration and validation layers between LLMs and external systems


• Developing verification and self-correction loops around runtime and tool failures


• Designing human-in-the-loop checkpoints for higher-risk autonomous actions


• Making sure agent infrastructure is scalable, resilient and secure

What I’m looking for:

• Strong software engineering or AI infrastructure background


• Experience with Docker / Kubernetes and ideally WASM or micro-VMs


• Strong API integration and systems engineering experience


• A good understanding of LLMs / generative AI / agentic systems


• A security-first mindset, particularly around prompt injection, data exfiltration and runtime security


• Someone who enjoys taking non-deterministic systems and building reliable engineering frameworks around them

You don’t need to have trained LLMs. I’m much more interested in someone who understands how to build the infrastructure around them and make them work safely at scale.