AI engineer in healthcare

We have a hugely exciting opportunity for a Senior AI Engineer who will take HealthHero's AI initiatives from prototype to production. Your first project will be our flagship, the Clinical Brain: a system that includes a Class II medical device, with an AI layer interacting with deterministic components. You will work on it directly with the Director of Clinical Products and the Head of Quality & Compliance, so you will need to be comfortable building to the rigour, safety, traceability and documentation that a regulated medical device environment requires.

• Partner with Product, Data, Engineering, Clinical and Operations teams to identify, scope and evaluate AI opportunities.
• Challenge requirements and technical approaches, assess feasibility, business value and costs, and help prioritise high-impact initiatives.
• Contribute to the AI roadmap, architectural decisions and technology choices, while helping define AI engineering best practices and sharing expertise across teams.
• Abstract patient intent: turn free-text and conversational patient input into structured, clinically meaningful intents.
• Condense and cluster questionnaires: reduce and group clinical questionnaires while preserving clinical meaning and safety.
• Combine NLP, LLMs and machine learning: design the AI layer and its interfaces with the deterministic components so that every output is traceable and explainable, with patient safety built in.
• Optimise the clinical graph: improve the structure and performance of the knowledge graph that underpins the Clinical Brain.
• Work with the Director of Clinical Products to translate clinical needs into AI behaviour, and to design and run clinical validation.
• Work with the Head of Quality & Compliance to meet Class II medical device requirements: design documentation, risk management, change control and traceability from requirement to test.
• Golden datasets: curate and maintain reference datasets with clinicians, covering common cases, edge cases and safety-critical scenarios.
• Mixtures of evaluators: go beyond a single LLM-as-judge by combining multiple judges, calibrating them against clinician review and measuring inter-rater agreement.
• Human-in-the-loop review: design efficient clinician review workflows that feed back into datasets, prompts and models.
• Regression testing: automate evaluation so that every prompt, model or data change is gated before release.
• Evidence for regulation: produce evaluation evidence that supports the Class II medical device documentation within the Clinical Brain, and monitor performance and safety once in production.
• AI Medical Note Taker: develop and integrate AI-powered clinical documentation into our consultation products, working with clinicians to ensure accuracy, reliability and clinical safety.
• Contact Centre AI (CCaaS): build and integrate call and chat summarisation, intent detection and routing, agent assist and self-service flows into our contact centre platform.
• AI-powered automation: design and deploy operational and clinical-admin workflows using tools such as n8n, with LLM steps, error handling, monitoring and clean handover to business teams.
• Knowledge graphs and RAG: develop knowledge graph capabilities using technologies such as Neo4j, Cypher and GraphRAG to ground LLM answers in trusted, structured data.
• AI agents and MCP: build secure Model Context Protocol services that enable AI agents to interact with healthcare systems, including searching appointment availability and booking teleconsultations, with authentication, consent, audit logging and safe failure modes.
• Self-hosted LLMs: benchmark open-weight models and serving stacks (e.g. vLLM, TGI) against hosted APIs on latency, throughput, cost and quality, and recommend where self-hosting makes sense.
• Product integration: integrate AI services into our TypeScript and .NET products and APIs alongside Product Engineering teams.
• Collaborate with Engineering and Data Platform teams to build scalable, reusable AI capabilities.
• Industrialise LLMOps, including prompt versioning, observability, cost tracking and quality monitoring.
• Build AI components fit for a Class II medical device environment: work within our quality management system and regulated software lifecycle, contribute to clinical risk management, and maintain design documentation and traceability.
• Embed privacy and data protection by design (GDPR), and support post-market monitoring of AI performance and safety in production.

• 4+ years of software engineering experience, including at least 2 years shipping LLM or ML-based features to production.
• Strong TypeScript (our codebase is mostly TypeScript and .NET), plus solid Python for AI and data work.
• Comfortable with APIs, cloud services (AWS, GCP or Azure), containers and CI/CD.
• Hands-on experience with LLM APIs (OpenAI, Anthropic, Mistral or similar) and frameworks for RAG, agents and tool use.
• Proven track record building evaluation pipelines (test sets, metrics, LLM-as-judge) and using the results to drive decisions, with experience of or strong interest in methods beyond a single judge, such as mixtures of evaluators, calibration against clinician labels and inter-rater agreement.
• Solid prompt engineering skills and a disciplined, experimental mindset.
• Experience with workflow automation tools such as n8n, Make or Zapier.
• Understanding of data privacy and security when working with sensitive personal data.
• Experience of, or a clear commitment to, working in a regulated medical device environment (Class II): documentation, change control, traceability and risk management.
• Clear communicator who can work with clinicians, product managers and non-technical stakeholders.
• Intent abstraction: classifying and abstracting what a patient needs from free-text or conversational input.
• Condensing and clustering questionnaires: reducing redundancy and grouping questions or responses while preserving clinical meaning.
• Hybrid NLP + LLM + machine learning systems designed for traceability, explainability and patient safety.
• Graph optimisation: graph algorithms, pathfinding or optimisation over clinical knowledge graphs.

• Generous holiday of up to 33 days per annum inclusive of bank holidays (pro-rata)
• Company Pension Scheme
• Simply Health cash plan membership
• Life Assurance