Context Engineer

About the role

The Context Engineer is a specialist software engineering role focused on ensuring that AI systems receive the right information, in the right structure, at the right time to produce reliable, high‑quality outputs.
While similar in seniority to a Software Engineering Specialist, this role differs fundamentally in skillset and ways of working. Instead of writing application logic or services, the Context Engineer designs and engineers the context layer that powers AI agents—system prompts, tool definitions, memory strategies, and retrieval pipelines—across the software development lifecycle (SDLC).
This role is critical to making AI‑enabled delivery predictable, scalable, and safe.

What you’ll be doing

1) Context & Prompt Engineering (Core)
• Design, implement, and maintain system prompts, developer prompts, and tool definitions that guide AI agents across different SDLC stages.
• Treat prompts and context artefacts as versioned, testable engineering assets, not ad‑hoc instructions.
• Ensure prompts are clear, precise, and aligned to desired outcomes, constraints, and quality standards.
2) Context Architecture & Schemas
• Define context schemas tailored to specific AI agents and SDLC phases (e.g. design agents, code‑generation agents, code‑review agents, test‑generation agents).
• Decide what information is required, optional, or excluded for each agent to maximise signal‑to‑noise ratio.
• Collaborate with architects and engineers to ensure context aligns with system boundaries, patterns, and guardrails.
3) Context Window & Memory Management
• Actively manage context window budgets, making informed decisions about:
o what content is retained
o what is summarised
o what is evicted as conversations or agent loops grow
• Design and maintain memory strategies (short‑term, long‑term, episodic, summarised) that support effective multi‑step reasoning.
4) Retrieval‑Augmented Generation (RAG)
• Design and maintain retrieval pipelines that surface relevant knowledge to AI agents at runtime.
• Own RAG components including:
o embeddings strategies
o vector databases
o re‑ranking and filtering logic
• Ensure retrieved context is accurate, relevant, and appropriately scoped for the task at hand.
5) Prompt Lifecycle, Testing & Optimisation
• Own prompt versioning, change control, and rollback strategies.
• Run A/B testing and evaluation of prompts and context strategies to measure impact on quality, speed, and reliability.
• Continuously refine prompts and context structures based on observed outcomes and failure modes.
6) Knowledge Codification & Collaboration
• Work closely with domain experts, engineers, and business stakeholders to codify institutional knowledge into AI‑consumable formats.
• Transform tacit knowledge, guidelines, standards, and playbooks into structured, retrievable context.
• Ensure knowledge remains current, governed, and reusable across teams and pods.
7) Governance, Risk & Responsible AI
• Ensure context and prompts embed appropriate constraints, safety rules, and quality expectations.
• Identify and mitigate risks such as hallucination, outdated knowledge, over‑broad context, or leakage of sensitive information.
• Support auditability by maintaining clear lineage of prompts, context sources, and changes.

Essential Skills / Experience

• Strong experience in prompt engineering and context design for LLM‑based systems.
• Solid understanding of information architecture and structuring complex knowledge.
• Hands‑on experience with RAG stacks, including vector databases, embeddings, and re‑ranking approaches.
• Deep understanding of token‑level reasoning, context windows, and model constraints.
• Exceptional written communication skills—clear, precise, and unambiguous writing is critical.
• Background in software engineering, data engineering, or platform engineering.
• Experience working with AI agents across different SDLC phases.
• Familiarity with evaluation frameworks for LLM outputs.
• Experience operating in regulated or enterprise environments.