Content Recommendation System: Built Content recommendation system to provide contextual content recommendations with dynamic display and native advertising integration for "What to read next" component of Yahoo's article pages. This large scale component is built in Java as serving layer, Storm for content processing and Redis as caching layer. This fully CI/CD component built using Docker and Kubernetes.
Dynamic Ad layout: Designed and implemented a system to get user engagement signals in real time and change the ad layout to optimize ad yield and revenue. It is built in Java with Kafka, Hbase and Storm.
Yahoo Recommends : Developed serving component for Yahoo Recommends - a content recommendation widget deployed on sites like CNET, Download.com, Hearst Publications, Vox Media
Label Classification: Built a system for editors to label the classifications and feed into active learning pipeline for content label classifiers.
App Install and Attribution: Developed serving component for delivering native app install promotion system on mobile web to increase yahoo in-house app installs and Daily Active Users (DAUs). Branch.io integration for app install attribution for existing app install promotions on Homepage Mobile Web Application.
Elections Result Backend System: Developed a reliable system to deliver near real time election results for our users across all devices.
Android: As a part of Yahoo's monetization team, worked on designing new ad formats for Android apps including full scroll ad unit, pencil ads. Designed and developed end-to-end solution for deep linking and deferred deep linking on Newsroom app.