Facebook Reels (October 2020 - December 2021)• Led the development of an ML based consumption experience feature by validating its feasibility through analyses, proposing to team, and collaborating with an engineer to build the prototype • Launched 8 improvements to Reels’ ML content ranking system, which reduced volume of user reports on content violating Trust & Safety policies by 66% while increasing Reels watch time by 5% Feed Ranking (July 2018 - December 2020)• Led team’s strategy for ranking reshared content to increase user conversations on Feed by proposing and executing a UX research analysis that identified the comment-efficient types of reshared content to promote to increase conversations• Launched the first ML ranking demotion targeting pirated movies on Facebook, which reduced their watch time by 20%, by sizing need for demotion, proposing to product team, and partnering with ML and Digital-Rights partners to build demotion• Launched an ML ranking demotion that reduced the prevalence of problematic content on the Feed by 45 million views/day by identifying the features used in the demotion through analyses and managing it's A/B tests• Built and implemented end-to-end tooling that enabled the Ranking team to monitor top viral content across key countries and debug the ranking rules unintentionally making bad content viral Photos Experience (January 2020 - December 2020)• Project managed the creation of an ML classification model that enabled the team to track, investigate, and fix image not loading issues reported by users at scale, resulting in a 20% decrease in reports and holding-the-line within guardrail goalsFacebook Summer Hackathon (July 2020)• Won Facebook hackathon by creating a feature that allows users to better manage and organize their workchat conversations

• Shipped localization bug fixes on the tablet video experience across key emerging markets by identifying markets through analyses and collaborating with international partners to conduct functionality tests abroad in such markets

• Saved Media Ops ~150 hours of manual classification work by ideating a web-scraping script that automated the manual process of collecting athletes’ social media accounts from public websites • Enabled Media Ops teams to more accurately categorize and track ~100M+ sports accounts by restructuring the sports classification taxonomy to be mutually exclusive and collectively exhaustive