Context Engineer

About the Role

As a Context Engineer at CapIntel, you'll sit at the intersection of software engineering and applied AI. This is a senior engineering role first, with a specialisation in integrating large language models into production systems. It is hands-on and production-focused rather than research-oriented: you'll be writing code in our core application roughly 75% of the time, and you'll own the features you build through to production support.

Our platform is built in Node and TypeScript, and you'll be working in it every day. You'll help define for how language models are integrated into that platform, and for how our engineering team adopts agentic workflows.

You'll be embedded in a development team working closely with engineers, product managers, and domain experts. As the first practitioner in this discipline at CapIntel, you'll also help define what context engineering looks like here, setting the patterns and practices the broader team can build on.

This role is ideal for a strong backend engineer who has already shipped customer-facing AI features, cares about production reliability over demo-day performance, and is energised by working in a discipline that's still taking shape.

What You'll Do

• Build and ship LLM-powered features in our Node/TypeScript application via model APIs (e.g. Anthropic, OpenAI, Bedrock), owning them from design through production support
• Help architect and maintain retrieval-augmented generation (RAG) pipelines, connecting language models to internal knowledge bases, databases, and live data sources, with accountability for retrieval quality as well as retrieval plumbing
• Collaborating with Architecture, manage context window strategy, determining what information enters the model, when, in what format, and at what level of compression to optimise for accuracy, cost, and latency
• Design and implement agentic workflows enabling the platform to handle multi-step, autonomous tasks, and diagnose them when they behave unexpectedly, including wrong tool selection, silent failures, and partial state
• Build guardrail and output validation layers that constrain model behaviour and ensure AI features act within well-defined, compliant boundaries
• Develop reusable agent primitives, prompt templates, and workflow components that other engineers can build on independently
• Build evaluation frameworks to measure context effectiveness, output quality, and agent reliability against real production traffic
• Monitor deployed AI systems for failure patterns and implement mitigation strategies, feeding learnings back into continuous improvement cycles
• Help define the engineering standards for this discipline at CapIntel, covering patterns, review criteria, testing approach, and documentation
• Collaborate with Product, Product Engineering, Implementation, and Data teams to translate business requirements and proofs of concept into production AI systems
• Act as an internal practitioner and resource, helping upskill the broader engineering team on context engineering principles and agentic best practices

What We're Looking For

• 5+ years of professional software engineering experience building and operating production systems
• Deep, current Node/TypeScript experience. This is a core requirement for the role, Python experience is a welcome addition alongside it
• 1 to 2+ years building LLM-backed features that external users depend on in production, rather than prototypes or internal tooling
• Working knowledge of RAG architecture, vector databases (e.g. Pinecone, pgvector, AWS OpenSearch), and semantic search, including an understanding of where retrieval quality degrades and why
• Hands-on experience with an orchestration or agent execution framework
• Familiarity with context management techniques: summarisation, chunking, session splitting, and memory strategies
• Experience debugging non-deterministic systems
• Experience introducing a new technical practice or discipline into a team that didn't have one, and getting other engineers to adopt it
• Experience building or consuming REST APIs and integrating with third-party services
• Comfortable collaborating with cross-functional teams in a fast-paced, high-growth environment, and making decisions with incomplete information as the field evolves
• Ability to communicate technical concepts clearly to both technical and non-technical partners

Nice to Have

• Experience in a regulated industry (financial services, healthcare, insurance) and awareness of what compliance requires of AI outputs
• Experience with the Model Context Protocol (MCP) or similar tool-integration standards
• Familiarity with LLMOps practices: tracing, observability (e.g. LangSmith, Langfuse, Datadog), model versioning, and rollback
• Exposure to multi-agent architectures and orchestration patterns
• Experience using AI-assisted development tools as an established part of your own workflow
• Familiarity with AWS or cloud-based infrastructure and containerised deployments (Docker, Kubernetes)

#LI-AC1