I sometimes build cool things that slightly work.

Architecting automated BI dashboards and ETL pipelines that turn chaotic student demographic data into something executive leadership can actually act on. Doing Propensity Score Matching to evaluate program outcomes and serving as data advisor to the DEIA committee.

Started here digging into student retention datasets with SQL and Python. Built ETL pipelines that took hours of manual work down to minutes and surfaced insights that contributed to a 200% improvement in retention rates.

Led a frontend team working on a Next.js/React app, set up TypeScript standards and CI/CD pipelines, and integrated Mixpanel + Google Analytics for data-driven A/B testing. Boosted conversion rates by 15% and cut technical debt by 30% by modernizing a legacy codebase. Also fixed more bugs than I'd like to admit.

Built a custom helpdesk dashboard connecting multiple third-party APIs through a Prisma/Postgres backend. Designed a smart-search feature with tag-based filtering that made digging through old tickets actually bearable. The 'find similar tickets' feature was the most useful thing I shipped that year. Allowing helpdesk to go through the mess of IT and understanding problems with limited context

Designed evolutionary algorithms for optimizing wildfire sensor node placement across Southern California terrain — accounting for topography, sensor range, and coverage overlap. Co-presented at SCCUR and was the main author at NCUR