• Ads Personalization - Designed and implemented systems to personalize ad components on the LinkedIn Audience Network using member attributes, aiming to enhance ad engagement and increase advertiser ROI.• Design, develop and scale systems responsible for decorating ads before they are displayed to a member or previewed by the advertisers.

• Developed a unified platform for hosting TensorFlow image and video machine learning models, leveraging ProML (Productive Machine Learning) to provide an end-to-end, unified solution for building machine learning systems at LinkedIn.• Collaborated closely with AI engineers to integrate and productionize various machine learning models, optimizing their thresholds and scaling them to actively filter inappropriate content on the site.

• Redesign user reporting flow to offer detailed reporting options for policy violations across key areas like Ads, Profiles, Feed, and Jobs, and update the UI to be consistent with the rest of the site, leading to an 18.6% reduction in user reports.• Introduce a new Disinterest modal to uplevel the actions users can take on uninteresting or objectionable content to deter overreporting and operationalize negative feedback signals in feed ranking.

Led efforts to comply with General Data Protection Regulation (GDPR) to avoid getting fined for 4% of company revenue by creating offline data deletion pipelines using Spark/Scala frameworks to delete member data from multiple third parties on account closure. This pipeline efficiently purges millions of records weekly inline with GDPR laws.

Implemented access layering for model publishing REST endpoint, ensuring secure authorization for AI teams accessing/publishing a model.

As a member of the Egencia Cars team, developed backend APIs to ensure that corporate clients can only select vehicles in accordance with their company's policies, such as budget and category.