Pushkar Tripathi
Distinguished Engineer at Meta focused on ads and recommendation systems
Summary
Work
Education
Projects
Writing
Sequence learning: A paradigm shift for personalized ads recommendations
November 1, 2024Engineering blog post describing Meta's move from human-engineered sparse features to event-based sequence learning for ads recommendations (co-authored in Meta engineering writing where Pushkar is listed among contributors).
Towards Fairness in Personalized Ads Using Impression Variance Aware Reinforcement Learning
January 1, 2023Co-authored work presenting a framework (Variance Reduction System) to reduce imbalances in ad impressions across demographic groups and reporting simulation and online A/B test results.
CoRTX: Contrastive Framework for Real-time Explanation
January 1, 2023Co-authored paper proposing a contrastive learning framework to train real-time explainers efficiently with limited explanation labels.
Allocation Problems with Partial Information
January 1, 2012Ph.D. dissertation addressing allocation problems under partial information; contains theory and algorithms related to stochastic/online allocation.
Matching with Our Eyes Closed
January 1, 2012Paper (co-authored) on a query-commit formulation for matching under uncertainty; addresses randomized greedy algorithms and approximation factors.