Data Science Projects

Experimentation Platform Enhancements

• Context: Inconsistent metrics across teams.
• Approach: Central metric definitions, guardrail metrics, CUPED variance reduction.
• Outcome: Faster reads on experiments; reduced false positives.

Marketing Mix Modeling Prototype

• Context: Need to allocate spend effectively.
• Approach: Hierarchical Bayesian MMM with saturation and adstock; feature engineering.
• Outcome: Actionable elasticity estimates; scenario planning tool.