• Science lead for ML-based prediction settlement model that provided propensity score/segmentation for marketing team’s event-based campaigns and loyalty campaigns to drive incremental ordered purchase sales and digital engagement on Amazon homepage• Designed and maintained ML pipeline for data integration, feature engineering, job monitoring, model training, performance evaluation, model selection, scheduled prediction and deliverables integration• Implemented AutoEncoder and TF-IDF to deal with high sparsity and dimensionality of customer data• Proposed local-to-global hybrid model framework for category/product granular level prediction