Mountain View, California, United States
❖ Collaborated with designers and data scientists to test how different presentations of Search features improve UX.
❖ Wrote technical design document for project, outlining deprecation, implementation, blockers, and A/B test plan.
❖ Led discussion with and leadership to gain their approval.
❖ Design protocol buffer that maps search result to feature proto. Implement frontend component using Google’s internal toolings to render the feature proto.
❖ Perform analysis on live experiment results with data scientists. Using the project - Replacing Appbar with Minibar for example, the live experiment covered 150 million unique users in 14 days.
❖ Experiment results showed no statistically significant changes in metrics, which is the hypothesis since the two components are similar in appearance.
❖ Enhanced the minibar by incorporating the new frontend soft-transitioning API, where navigating to the next query via Minibar tiles rebuilds the page on the client-side with protocol buffer from backend. Live experiment shows this enhancement reduced time to first click (ttfc) of the next query decreased by 3.00%
❖ Enabled Minibar to have a sort control when the data contains dates. Reduced manual refinement (manref) by 2.32% and further reduced time to first click (ttfc) by 1.59%.