Data Scientist, Data Quality & Provenance

In this role, you will define rigorous data-quality concepts and build metrics that prove the trustworthiness of data, annotations, and evaluation signals for our embodied AI systems. You’ll translate practical quality questions into defensible metrics and deliver automated reports that communicate confidence and limitations. Working with annotation, autonomy, and evaluation teams, you’ll analyze large datasets with Python and SQL to steer training data, evaluation, and system improvements. This is an opportunity to shape data-quality standards across a safety-critical, platform-first AI company.

• salary benchmarked against market
• meaningful equity
• relocation and visa support
• hybrid working with core hours
• learning and development budgets
• comprehensive benefits (health, dental, enhanced leave, pension, wellbeing support)

• Define measurable, statistically rigorous data-quality concepts (coverage, label quality, uncertainty, provenance, model performance)
• Apply statistical inference to multi-rater annotation, label ambiguity, dataset coverage, and black-box model evaluation
• Collaborate with annotation, autonomy and evaluation teams to translate practical quality questions into defensible metrics
• Build automated reports that clearly communicate confidence, limitations and interpretation
• Analyse large annotation datasets using Python and SQL to identify quality issues and inform decisions
• Iterate on metrics and reporting based on user feedback
• Develop reusable quality artefacts to replace ad-hoc analyses with automated metrics and reports
• Work on multi-rater and human-in-the-loop ML systems, including Dawid–Skene-like models, psychometrics and inter-rater agreement
• Define quality standards spanning coverage, uncertainty, provenance, annotation quality and predictive-model performance
• Integrate data-science outputs into platform workflows used by engineering and research teams

• strong foundations in classical statistics (experimental design, sampling, modelling, estimation, inference)
• practical experience with rater/annotator modelling, inter-rater agreement, label uncertainty or related methods
• understanding of data-quality dimensions (coverage, label quality, uncertainty, provenance, model performance)
• experience with annotation tooling, QA methodologies, human-in-the-loop ML, or multi-rater labelled datasets
• strong Python and SQL skills with ability to analyse large, complex datasets
• ability to translate ambiguous stakeholder questions into rigorous, usable metrics and communicate uncertainty and implications clearly
• advantageous experience with probabilistic or Bayesian modelling, production platform integrations or safety-critical data

• clear communication of uncertainty and limitations
• ability to translate stakeholder questions into measurable metrics
• collaborative with cross-functional teams
• Python
• SQL
• statistical methods for data quality