At Vellum, I analyzed onboarding behavior for 12,000+ users in SQL and BigQuery, identifying drop-offs that informed a redesign and lifted completion from 24% to 77%. I also built cohort retention dashboards for 27,000+ users as D1 retention rose from 5% to 18.5% across launches.
I checked growth KPI data and removed 2,700+ bot signups, representing 13% of launch-month volume. Using LogRocket session replays, I helped prioritize UX fixes, and my analysis of six acquisition channels found that AI search referrals converted at about nine times the site average.
At Chapman University NSF-REU, I analyzed 20 years of NASA Giovanni satellite data to study cyclone formation across the Arabian Sea and Bay of Bengal. I cleaned and transformed the data in Excel and compared regional time series, including observations around the September 2005 cyclone.
In my projects, I built a DuckDB and dbt pipeline that turns 99,000+ raw orders into 16 analytics models, with 44 automated data tests. I’ve also modeled NBA game outcomes and player aging, including research showing that a standard method overstates late-career decline by 45%.
