Product Data Scientist - Learning Platforms Infinitas Technology

Purpose The Product Data Scientist raises the standard of how we measure and prove impact across our digital learning platform. You set the bar for trustworthy metrics, valid comparisons, and sound evidence and use it to answer the questions that shape our product. The scope covers learner and teacher experiences across our Learning Platforms. Becoming genuinely data-driven is a core pillar for our company, and this role is important in reaching that goal.

Key responsibilities Experimentation and causal evaluation Design A/B tests and, where controlled testing isn't possible (seasonality, classroom-locked cohorts, staged rollouts), apply quasi-experimental methods such as difference-in-differences, matching, and interrupted time series. Power analysis and minimum detectable effect up front; disciplined handling of peeking, multiple comparisons, and variance reduction.

Statistical modelling for validation and insight Look beyond single metrics and simple comparisons. Use models to find what really drives an outcome, to account for the fact that users sit within groups such as classes and schools, to validate our data, and to test assumptions with forward-looking estimates. This is modelling to understand and validate not to build live machine learning.

Learning outcomes and efficacy Build the evidence base that learners make progress on our platforms. Design A/B tests and comparison-group analyses that test whether product changes improve learning, accounting for prior ability and teacher and school effects. Success metrics: design and validation Define success metrics and guardrails for product initiatives, and validate them, does the metric measure what we claim, is it sensitive enough to detect real change, does it hold up over time and across segments.

Own the definitions that everyone else builds on. Analytical standards and judgement Set the standards for how analysis is done here: unit of analysis, when numbers may be aggregated and when they may not, weighting, comparability across products and opcos, and how uncertainty is communicated. Review the work of others and raise the bar through that review.

AI-assisted analysis and analytical agents Work with engineers to build and evaluate agents that support analysis (automated experiment readouts, anomaly detection, querying over certified data models). Own the evaluation side and analysis guardrails: define how we know an agent's answer is correct before we trust it at scale. Measurement infrastructure and data quality Co-own tracking and instrumentation plans with Engineering.

Ensure the metrics we depend on are accurate, documented, monitored, and traceable. Communication and decision impact Turn analysis into short, sharp narratives with a clear recommendation and honest trade-offs. Be equally willing to say what the data supports and what it cannot answer.