About this role
As an Applied Scientist II specializing in lead scoring and deep learning modeling, you will build and improve machine learning models that power how our business engages with customers. You will develop predictive models for customer segmentation, scoring, and lead/account prioritization, working within an established scoring architecture and collaborating with senior scientists and cross-functional teams to deliver production-grade components. Key job responsibilities * Build and iterate on predictive lead scoring models to support customer acquisition, conversion, and retention strategies using techniques such as survival analysis, graph networks, or transformer-based architectures. * Develop and maintain ML pipeline components for deep learning models, including data preprocessing, feature engineering, model training, and inference integration. * Contribute to internal and external research, including science reviews, technical publications, and patent filings in collaboration with senior scientists. * Apply multi-modal modeling techniques (text, graph, behavioral, and temporal data) to enhance scoring accuracy across account and lead levels. * Conduct A/B testing, causal inference, and counterfactual analysis to measure model impact and iterate on model design. * Partner with MLOps engineers on model deployment, monitoring, and retraining using tools like AWS SageMaker, MLflow, and other internal tools. * Participate in science reviews to maintain and raise the quality bar within the team. * Implement and execute offline and online evaluation frameworks; track success metrics tied to business outcomes (conversion rates, pipeline generation). About the team The AWS Marketing Science team builds the ML models and measurement systems that drive marketing decisions across Amazon Web Services. We own incrementality and valuation, ROI measurement, marketing attribution, propensity scoring, account and lead clustering, and next-best-action models. Our work directly influences how AWS allocates marketing spend, targets accounts, and measures effectiveness across billions in pipeline.