From Statistical Research to Experimentation Capabilities

I led research and platform adoption of CUPED and rerandomization methods to improve the sensitivity and reliability of large-scale online experiments.

The problem

Reliable experimentation requires effective experimental design and valid inference. Covariate imbalance and variance-reduction choices affect how precisely effects can be estimated and how results should be interpreted.

My role and approach

Led work on the theoretical foundations and validation of updated CUPED methods. Led rerandomization research and product design, partnering with product and engineering teams on implementation and adoption. Established a framework combining theoretical analysis, simulation, and historical-data validation.

From methods to practice

This work connected statistical research with product workflows, validation, and platform adoption. CUPED and rerandomization address complementary parts of the experimentation process: analysis and design.

Public research

Bridging Control Variates and Regression Adjustment in A/B Testing — arXiv preprint, 2025. https://arxiv.org/abs/2509.13944v2

Ensuring Trustworthy Online A/B Testing: Addressing Five Key Questions on CUPED — arXiv preprint, 2026. https://arxiv.org/abs/2606.18750v1