Lead Data Scientist

Role Overview

The Lead Data Scientist in the Research & Rapid Prototyping team owns the research agenda for a problem area within our eDiscovery platform, translating ambiguous business needs into tractable research questions and driving them through to working prototypes.

This is a senior individual contributor role with significant technical influence. You will set the direction and the standard for how we apply Large Language Models and other advanced techniques to enterprise and legal documents, and you will mentor other data scientists on the team as you do it. The role is a natural stepping stone toward people leadership: you will develop and guide others without carrying formal line management responsibility.

Responsibilities

• Technical Leadership: Own end-to-end delivery of research workstreams from problem framing and scoping, through experimentation and prototyping, to handoff to engineering and product. Make and defend the technical calls that shape how a problem gets solved.
• LLM Specialisation: Set the direction for how we implement, fine-tune, and deploy local Large Language Models for eDiscovery, with a focus on their applicability and efficiency in processing and analysing legal documents such as emails and PDFs.
• Evaluation Methodology: Own the evaluation approach the team works to. Define what "good" looks like for legal-specific tasks, build the benchmarks that measure it, and ensure results are trustworthy enough to make decisions on.
• Mentorship & Team Development: Mentor junior and senior data scientists through experimental design reviews, code review, and pairing. Raise the bar on research rigour across the team and help others grow into more ambiguous work.
• Horizon Scanning: Evaluate emerging models, tools, and techniques, and make clear build/buy/adopt recommendations. Decide what merits investment, and what should be left alone for now.
• Rapid Prototyping: Quickly build prototypes that demonstrate how emerging technologies can be applied in practice, exercising the judgement to know when a prototype should move to production or whether to move on.
• Statistical Analysis & NLP: Apply statistical analysis and traditional NLP techniques alongside modern methods to extract insights and deepen our understanding of enterprise and legal documents.
• Stakeholder Influence: Communicate findings, limitations, and trade-offs clearly to non-technical leadership. Influence the product roadmap based on what is genuinely technically feasible.
• Multi-modal & Graph Technologies: (Nice to have) Work with multi-modal data (images, audio, etc.) and explore the integration of Knowledge Graphs and other emerging technologies to enrich our solutions.

What You'll Bring

• Experience: 58 years of applied data science or machine learning experience, including demonstrable depth in modern NLP.

• Education: Bachelor's degree in Computer Science, Data Science, or a related quantitative discipline. Master's preferred.

• Modern NLP Depth: Strong background working with LLMs, statistical analysis, traditional NLP, and enterprise documents, particularly legal and enterprise documents.

• Technical Stack: Hands-on experience with locally hosted and open-weight LLMs, and a working command of tooling such as:

• Serving & inference: vLLM and a working knowledge of deployment infrastructure (Kubernetes, containerisation etc)

• Modelling: PyTorch, Huggingface transformers, PEFT/LoRA fine-tuning

• Orchestration: LangGraph or comparable agentic/workflow frameworks

• Retrieval: embedding models and vector stores (pgvector or similar)

• Traditional NLP & ML: spaCy, scikit-learn

• Experimentation & delivery: experiment tracking (dvc or similar), Docker, Git

• Track Record: Evidence of research that reached production, or published, patented, or otherwise externally recognised work.

• Mentorship: Demonstrated experience mentoring or coaching other technical practitioners — not just willingness to do so.

• Ambiguity Navigation: Able to define the problem, not only solve one that has been handed over. Comfortable initiating work and driving it forward with minimal guidance.

• Language skills: Ability and patience to work with large amounts of natural language data with attention to detail required in the legal domain.

• Communication: Excellent verbal and written communication, with a knack for asking insightful questions, openly admitting knowledge gaps, and collaborating effectively.

• Multi-Disciplinary Collaboration: Able to work closely with multi-disciplinary professionals, both within and outside the Research & Rapid Prototyping team, to complete projects and achieve goals.

• Learning Agility: Eagerness to learn about the legal domain in order to better understand the data and its complexities.

• Experimental Mindset: Comfortable with trial and error, willing to admit and learn from failures, and unafraid to try new approaches.

Work Mode

Hybrid — 3 days per week working from the office.