Job reference: 001285
Salary: Competitive Day Rate
Location: All UK Regional Centres
Employment type: Contract
Directorate: 22 - CDIO - Core
Closing date: 26/11/2026

AI/Agent Engineer

6-12 month contract (dependent on business need)

Any HMRC Regional Centre across the United Kingdom

Competitive Day Rate

Role overview

Build and test AI prototypes, agents and reusable components; configure models, retrieval and evaluation tooling; and document and transfer technical assets so that successful propositions can be adopted and scaled.

Job description

You will work in a multidisciplinary AI Office team to turn priority use cases into working prototypes and reusable technical assets. You will design, build and test agentic and generative-AI solutions, integrating models with approved knowledge sources, tools and services while applying secure software-engineering practices.

The focus is on rapid learning without sacrificing engineering quality. You will create reproducible code, automated tests, evaluation evidence and technical documentation, then transfer assets and knowledge to the teams that will own, operate or further develop the solution.

  • Design, build and iterate AI prototypes, assistants and agents using approved models, frameworks and cloud services.
  • Develop reusable components for prompting, orchestration, tool use, memory, guardrails, identity and access control.
  • Configure retrieval pipelines, embeddings, vector search, chunking, ranking and grounding against trusted knowledge sources.
  • Create automated and human evaluation approaches covering quality, groundedness, safety, latency, reliability and cost.
  • Integrate agents with APIs and operational systems using secure, maintainable software patterns.
  • Write clean, tested and version-controlled code and establish continuous integration and delivery practices appropriate to the proposition.
  • Instrument prototypes and services for logging, tracing, metrics, model and prompt performance, and operational diagnosis.
  • Investigate failures, hallucinations, prompt-injection risks and other unsafe behaviours, implementing proportionate mitigations.
  • Produce technical documentation, deployment instructions, decision records and reusable examples.
  • Transfer code, evidence and knowledge to receiving teams, supporting handover and future development.

Person specification

A practical senior engineer who can work effectively with uncertainty and convert concepts into safe, testable software. You will be able to select appropriate tools, explain engineering trade-offs, collaborate with architects and subject-matter experts, and coach others in modern AI engineering practices.

Essential criteria

The successful contractor will be able to demonstrate the following essential experience and capability:

  • Experience in AI engineering.
  • Strong software engineering experience using Python and/or another modern language to build tested, maintainable services and APIs.
  • Hands-on experience building solutions with large language models, including prompt design, structured outputs, tool/function calling or agent orchestration.
  • Experience implementing retrieval-augmented generation, including ingestion, embeddings, vector search, chunking and relevance improvement.
  • Experience designing and running systematic evaluations for generative-AI solutions, with evidence of using results to improve quality and safety.
  • Knowledge of secure development, identity, secrets management, data protection and mitigations for AI-specific threats such as prompt injection.
  • Experience with source control, automated testing, continuous integration and deployment, observability and troubleshooting.
  • Ability to document and communicate complex technical work and transfer reusable assets and knowledge to other engineering teams.