Built and maintained tools in PHP (Hack), JavaScript, and ReactJS used by a large contingent workforce (in-house and crowdsourced) to provide millions of points of relevance and classification data daily to Facebook’s product teams including those focused on search ranking, news, places, videos, trending topics, NLP, and others.

Designed and built internal tools with PHP and SQL and developed metrics for internal teams and vendors to track data quality including visualizers for accuracy, inter-rater reliability, confusion matrices, precision / recall, and disagreements.

Collaborated with a team of software engineers as a member of a 24/7 on-call rotation; including winning a “fix of the month” award for quickly spinning up a framework to prevent user privacy policy violations in our human rating platform.

Designed and implemented a logging and reporting system based on an internal logging API, a Python and Apache Hive-based ETL pipeline, and a PHP-based vendor-facing tool to automate time-tracking and billing for contingent workers saving millions of dollars in lost productivity arising from inaccurate/exaggerated time reports and slow task processing.

Managed and developed training programs for a team of 40+ onsite contingent workers, overseeing cross-functional relationships with vendors and other teams as our rating workforce scaled from around 30 onsite workers to 2000+ human raters globally.