Agentic AI Engineer (life sciences) Cambridge

Key responsibilities

• Build and refine AI-agentic workflows on our internal framework to help drug discovery scientists generate the best testable therapeutic hypotheses to progress.
• Translate discovery problems into agent designs, working out where an agentic approach genuinely adds value, and where it doesn’t
• Integrate agents with our knowledge graph, proprietary methods, scientific literature and other sources so they reason over the right evidence for each problem
• Expand and maintain our existing genAI tools so they keep pace with the team’s needs and a fast-moving ecosystem
• Contribute to how we evaluate agentic workflows, helping shape sensible evals, testing and quality standards as the practice matures
• Write clear, maintainable, well-documented code that others on the team can build on

What success looks like

In 3 months, you have:

• Got to grips with our agentic framework and the discovery problems it serves, and made your first contributions land in the codebase
• Built working relationships with the scientists and engineers you partner with, and started turning their feedback into concrete improvements

In 6 months you have:

• Taken at least one agentic workflow from idea to a production tool scientists use in our drug discovery pipeline – built on our framework, with sensible evaluation and documentation.
• Operating with real independence – owning agentic workflows end to end, proposing improvements

What we are looking for

We’d love to hear from you if your experience and interests overlap with much of this profile (and we don’t expect every box ticked – if you’ve built things you’re proud of and you’re excited about what Healx does, get in touch):

• You’ve built and shipped LLM-agentic systems that deliver real value — agents that use tools, orchestrate multi-step workflows and behave reliably in production. We care more about what you’ve built than how long you’ve been at it.
• Strong software engineering fundamentals — you write clear, tested, maintainable code that others can build on
• Fluency with the modern LLM/agent toolkit — model APIs, prompting, tool use, RAG, and the patterns this fast-moving ecosystem is converging on (MCP, agent frameworks, evals).
• A pragmatic sense of where an agentic approach genuinely helps and where it doesn’t, and you enjoy working closely with non-engineers — the kind of person who’ll sit with a scientist to understand what they actually need.

It’s a bonus if you have:

• Experience in drug discovery, biology, or another life scientific domain — or simply a real appetite to learn one
• Familiarity with knowledge graphs or reasoning over structured or heterogeneous data
• Experience building evaluation harnesses or testing strategies for LLM systems
• A track record of picking up unfamiliar domains quickly and becoming useful fast
• Interest in or experience with biotech / techbio and its impact on patient outcomes

#J-18808-Ljbffr…