AI Software Engineer | Python | RAG | Retrieval Augmented Generation | DAG | Dagster | London, UK

The role involves hiring an agent-focused software engineer to build internal agentic frameworks for our discovery product. You will define and implement the operating system that allows scientists to run repeatable, agent-in-the-loop, tool-using workflows while maintaining strong provenance, traceability, and review checkpoints.

Initial priorities:

• Design and implement the core orchestration layer for scientist-in-the-loop, agent-based discovery workflows that integrate with the existing data, model, and workflow stack.
• Build a robust tool registry and execution framework so agents or scientists can safely invoke internal tools (e.g. model inference, dataset retrieval and quality control, literature search, bioinformatics workflows) with clear inputs and outputs.
• Establish provenance, logging, and evaluation infrastructure so every agent run is reproducible and reviewable by scientists and customers.

Core responsibilities:

• Own the architecture and implementation of agentic workflows (Python-first), including state management, retries, branching logic, and human-in-the-loop checkpoints.
• Build and maintain a tooling interface layer that standardises how agents call internal services and external APIs, with strong typing, validation, and error handling.
• Implement end-to-end provenance and traceability, including versioned inputs, prompt and tool versions, dataset and model references, and run-level audit trails.
• Create evaluation frameworks for agent and scientist performance (e.g. correctness, evidence coverage, precedent accuracy) and feed learnings from real projects back into system improvements.
• Partner closely with machine-learning, data, and scientific teams to expose models and data as agent-safe tools and ensure workflows reflect real discovery practice.
• Contribute directly to near-term customer deliverables by shipping a minimum viable discovery workflow aligned with the long-term platform architecture.

Additional responsibilities:

• Help shape engineering standards for agent reliability, safety, and interpretability in a scientific context.
• Support internal documentation, developer experience, and onboarding as the agent platform becomes shared infrastructure across the company.
• Potentially mentor future hires in agent engineering, orchestration, and platform development.

Core competencies:

• Strong software engineering background (industry or research), with deep experience building production-grade Python systems.
• Proven experience designing and deploying workflow or orchestration systems (e.g. DAGs, event-driven services) in complex domains.
• Experience working in scientific, biotech, or other high-integrity environments where reproducibility and auditability are critical.
• Hands-on experience working close to model APIs while maintaining clean abstraction boundaries.
• Strong systems thinking, balancing speed with long-term architecture and designing modular interfaces that prevent tooling sprawl.
• Experience implementing logging, observability, and evaluation for ML or AI systems.
• Ability to communicate clearly across disciplines and translate real scientific workflows into robust software.

Nice-to-have experience:

• Experience with agent frameworks, retrieval-augmented generation, or multi-agent systems.
• Familiarity with ML experiment tracking or model registries and data orchestration platforms.
• Exposure to knowledge-graph or evidence-graph representations and structured scientific reporting.
• Interest in plant biology, gene regulation, or crop improvement (not required).

Benefits:

• Competitive salary and equity options.
• Generous annual leave and flexible working policies.
• Benefits package and career development opportunities as the company scales.
• Ownership of ambitious, mission-driven work with real-world impact.
• Supportive, innovative team environment with access to conferences, events, and professional development resources.