Prompt Engineer (Copilot & Agent Building)

Location: London / Hybrid

Contract Type: 12-Month Fixed Term Contract

Working Pattern: Full-time

Clearance: BPSS eligible

We’re looking for an experienced Prompt Engineer to join a major UK government project focused on accelerating the safe and effective use of AI across departments. You’ll design, test, and optimise prompts and AI agents that power Microsoft Copilot and other large language model applications—helping teams automate routine tasks, enhance workflows, and make better use of AI capabilities in a secure, responsible way.

What you’ll do

• Design and optimise prompts for Copilot and other AI agents to ensure accuracy, safety, and contextual relevance.
• Build and refine autonomous AI agents using frameworks and decision-making algorithms.
• Integrate AI agents into tools, APIs, or existing government workflows.
• Work closely with researchers, engineers, and product teams to test and enhance agent performance.
• Develop evaluation frameworks, feedback loops, and safety mechanisms to monitor reliability.
• Maintain clear documentation of prompt libraries, design decisions, and best practice guidance.

What we’re looking for

• Strong experience designing and refining prompts for LLMs, copilots, or AI agents.
• Solid understanding of AI-driven systems, model behaviours, and safe deployment practices.
• Experience working in complex or government environments, ideally involving Microsoft Copilot or Microsoft 365 integrations.
• Skilled at analysing conversational logs and model outputs to identify improvements.
• Confident communicator who can explain technical concepts to non-technical stakeholders.
• Awareness of the ethical and safety considerations in deploying AI systems.

Desirable:

• Experience with agentic AI frameworks (e.g. AutoGPT, CrewAI, LangGraph, ReAct).
• Background in conversational design, human-AI interaction, or reinforcement learning.
• Familiarity with RAG, vector databases, or simulation environments.