We're looking for a Principal Data Scientist to lead the design of AI systems that can support complex, multi-step tasks, use tools, adapt to context, and make intelligent decisions. You'll work closely with product and engineering to architect LLM-powered features that are robust, useful, and production-ready not just impressive demos. This role is ideal for someone who not only understands data and AI deeply but also brings strong software engineering instincts. Someone who can design intelligent systems that operate reliably and scalably within a complex product environment.
What you'll be doing

• You will explore and implement RAG or Graph RAG approaches to improve retrieval quality and reasoning in LLM workflows, including graph construction, entity linking, and hybrid scoring strategies.
• Design and implement LLM-powered systems that support multi-step task automation, tool use, and context-aware reasoning within the product.
• Contribute to prompt design and evaluate prompts results
• Design and implement MLOps pipelines with best practises in mind
• Prototype, evaluate, and ship AI features using platforms like Claude or Amazon Bedrock with a focus on production readiness, not just experimentation.
• Develop evaluation frameworks for testing model output quality, reliability, and alignment with user goals (e.g., hallucination detection, prompt regression, safety scoring).
• Work closely with engineers and designers to embed AI capabilities into real user workflows, ensuring that technical architecture supports meaningful product outcomes.
• Own the full lifecycle of AI components from evaluation frameworks, planning and modeling to testing, deployment, and continuous improvement.
• Contribute to our AI strategy and roadmap, helping shape how we scale LLM usage responsibly across the platform. You bridge product and engineering by embedding AI into user-facing features, not just prototypes.
• Collaborate closely with Principal Engineer to align AI workflows, system prompts, and model interaction layers.

• You're a practical AI builder. Not just a theorist. You think in terms of customer value, vs. research or academic novelty.
• You've worked with Claude and/or Amazon Bedrock (or similar LLM platforms) and understand prompt design, model behavior tuning, and output evaluation.
• You're comfortable writing code, reviewing PR's, and deliver a reliable, explainable, production-ready product.
• Are excited about MLOps and brings experience in implementing MLOps best practices.
• You're curious about how people interact with AI and want to build systems they can trust and understand.

• Experience fine tuning models.
• Ability to reason about multi-hop question answering, graph traversal, and integrating structured retrieval with LLM prompts.
• Performance over time.
• 5+ years experience in applied AI, machine learning systems, or data science roles with production responsibility.
• Proficiency in Python, with experience building and maintaining AI systems, not just analyses.
• Experience working in Java or TypeScript environments beneficial.
• Experience with Claude, OpenAI, Bedrock is nice to have or experience with similar LLM platforms required.
• Familiarity with RAG, Graph-based retrieval, prompt design, and multi-hop reasoning.
• Experience deploying LLM-powered features into production environments.
• Bonus: experience in vector stores, agent orchestration, or legaltech domain knowledge.

• Contributory pension plan.
• 26 days annual holidays, hybrid working, and length of service entitlement.
• Health Insurance.
• Loyalty Share Scheme.
• Enhanced Maternity and Paternity.
• Employee Assistance Programme.
• Electric Vehicle Salary Sacrifice.
• Cycle to Work Scheme.
• Calm and Mindfulness sessions.
• A day of leave to volunteer for charity or dependent cover.
• Accessible and modern office space and regular company social events.