Lead Data Scientist

Role Purpose

The Lead Data Scientist is responsible for leading the development and refinement of advanced statistical models and machine learning techniques, creating innovative metrics and diverse insights that enhance the analysis of player and team performance. They will be exploring new methodologies in football analytics and continuously pushing the boundaries of data science to provide the club with cutting‑edge research. The department’s mission is to understand football better than our rivals and turn that understanding into decisions that win matches and trophies by building and maintaining a single source of truth, developing innovative quantitative models, and delivering tools and processes that embed timely, actionable insight into the Club’s critical football decisions.

The role would ideally be based at the club’s world‑class Training Centre in North London but we would be open to remote working for the right candidate.

• Develop advanced statistical and machine learning models to generate innovative metrics and insights from varied and complex data sources, enhancing the understanding of player and team performance.
• Manage the full modelling lifecycle, ensuring best practices in data science, statistical rigour and engineering are applied from data collection through to model validation, automation and delivery to the end users.
• Undertake research projects that address critical football questions, delivering novel insights that shape club strategy and decision‑making.
• Work closely with Insights Analysts across scouting and performance areas to translate data research into practical, actionable use cases.
• Integrate with Data Engineers to leverage and evolve a modern tech stack, ensuring the seamless merging of data analytics capabilities with the club's broader technological infrastructure.
• Collaborate with domain experts to understand football technical and tactical concepts and translate them into measurable metrics.
• Present your findings to a wide range of audiences across the club.
• Uphold our technical best practices and standards and mentor staff on good programming practices and coding literacy.
• Follow your curiosity to investigate unexplored research projects that develop our understanding of the game and provide a competitive advantage to the club.

• Master’s or higher degree in a quantitative field (Mathematics, Statistics, Computer Science, or related fields).
• Demonstrable experience as a Data Scientist or combined experience with a related role within data science or research.
• Extensive technical experience in applying data science, statistical modelling, and machine learning techniques.
• Expert understanding of statistical programming languages, especially Python and/or R, for data analysis and model development.
• Proficient in SQL development and knowledgeable about database and data warehousing technologies.
• Experienced with cloud‑based computing environments, ideally GCP, demonstrating proficiency in utilising cloud resources for data analytics and ensuring seamless integration of data pipelines and analytics platforms.
• Skilled with cloud‑based machine learning toolsets, such as PyTorch or equivalent ML frameworks, to leverage state‑of‑the‑art cloud technologies for advanced analytical modelling.
• Understanding of CI/CD practices, including the use of repositories for versioning and control, to ensure robust, scalable and reproducible code management within data science projects.
• Excellent presentation and communication skills, capable of effectively presenting and simplifying complex insights to a diverse range of stakeholders.
• A track record of conducting innovative research, applied within professional sports settings, or disseminated through publications, industry forums, or social media.

• Understanding of data architecture principles, including the design and implementation of robust data models and ETL processes, to support the construction of scalable and efficient data systems.
• Familiarity with big data tools like DuckDB, Polars or Apache Spark for processing "out‑of‑memory" datasets.
• Proven experience…