Agentic Systems Engineer (Digital Services)

Overview We're looking for an Agentic Systems Engineer, to work on systems that haven't existed before. Think of an Agentic Coach that navigates the unpredictability of human biology, or the software architecture behind a fully automated Dark Lab. These aren't chatbots waiting for questions. They're proactive systems that maintain world models, interpret human intent, and intervene before the user knows to ask. These systems create genuine impact for clients tackling complex problems at the forefront of technology. Building them means designing and wiring agent harnesses: layered architectures that span from sensor-edge reasoning to high-level planning, backed by stateful multi-turn reasoning, persistent per-user memory, and robust evaluation pipelines. You'll work in two directions at once. Internally, you'll use the latest AI tools to sharpen how we build, prototype, and ship. Externally, you'll deliver the breakthrough systems our clients deploy, designed to move from prototype to production at pace. There's no single technology stack. You'll work with agents, multi-agent systems, and the latest generative AI tools, but your work will always be grounded in solid software engineering. Expect to build scalable distributed systems in core languages like Python and TypeScript, leveraging frameworks such as NextJS and Pydantic, agent orchestration tools (LangGraph, CrewAI), and protocols like MCP. You'll deploy on cloud platforms (AWS, GCP, Azure) using Docker, Kubernetes, and infrastructure-as-code, backed by rigorous CI/CD and observability (OpenTelemetry, Prometheus). Due to the clients we work with, you must be eligible for UK security clearance. What you'll be doing: AI System Architecture: Designing and implementing the logic for autonomous systems, including maintaining world models, integrating multimodal sensing, and building real-time human-machine interaction loops. Infrastructure & Production Engineering: Designing the underlying software for scalable, secure, and observable cloud deployments. You will apply the engineering rigor required to take experimental AI from concept to robust live launch. Rapid Client Prototyping: Driving a demo-first delivery model. You will focus on fast experimentation to actively show tangible value and working software to clients, while maintaining the engineering standards required for production deployment. Domain Translation & Integration: Partnering directly with deep-tech subject matter experts (like biologists and roboticists) to apply software solutions to complex physical constraints. Agent Evaluation & Optimisation: Developing frameworks to measure, monitor and improve agent performance. You will design evaluation strategies, identify failure modes, and apply systematic testing to ensure agents are reliable, predictable and effective in real-world environments. What you need to succeed: Applied AI Expertise: You have hands-on experience building LLM-based agentic systems, with a solid grasp of the fundamentals: planning, reasoning, tool use, memory and orchestration. You know how to evaluate and monitor agents so you can measure how they perform in production. Core Software Fluency: You possess strong programming skills, good testing and deployment practices, and practical experience in cloud engineering, containerisation, and modern DevOps. You can design scalable, resilient systems, and you know that great agents still require excellent software engineering foundations. Navigating Ambiguity: You are comfortable working with uncertainty and evolving requirements. You break complex, ill-defined problems into clear, logical steps. You recognise that requirements evolve as new information emerges and can make sound engineering judgements about when rapid experimentation is appropriate and when a more rigorous approach is required. Communication: You communicate complex technical concepts clearly and confidently, adapting your style to diff