Data science & experimentation
I build models and experiments that explain and predict what users do — and translate them into decisions for product and business teams.
Selected work
Retention prediction (consulting, 2023–2025). Led analytics workstreams for a long-term client: identified behavioral markers of persistent users across 10 years of product data and built predictive models that improved retention-prediction performance by 23%, then turned the results into retention recommendations for non-technical stakeholders.
Behavior at platform scale. Modeled the longitudinal behavior of millions of users on brain-training and online-learning platforms, and of players in real-world tennis competitions, in repeatable R/Python pipelines.
Experiment design. Designed and ran online studies with daily data-quality checks, preregistration and open data, and analyzed them with Bayesian (Stan/MCMC) and mixed-effects models.
Toolkit: R, Python, SQL, Stan, Jupyter, Git, Power BI, Tableau · A/B testing and experimental design · regression, mixed-effects and Bayesian models · predictive modeling and machine learning · data visualization.