The Customer Behavior Analytics team designs innovative machine learning solutions to enhance customer experiences and strengthen their relationship with Amazon. This interdisciplinary team of scientists and engineers incubates and develops disruptive solutions using state-of-the-art technology to tackle some of the most challenging scientific problems in customer behavior analysis at Amazon. To achieve this, the team utilizes methods from deep learning, large language models (LLMs), natural language models, recommendation systems, affinity models, reinforcement learning, and econometrics to drive personalized experiences throughout the customer journey.

- Design and fine-tune language and generative models for recommendation and engagement, including continued pre-training, supervised fine-tuning, and preference-based alignment, to optimize for long-term customer value rather than short-term clicks.

- Develop generative recommendation and decision models that produce next-best customer engagement actions (e.g., recommendations, bundles, messaging, incentives, timing), conditioned on structured customer and household-level behavioral context.

- Build structured, temporal representations of customer behavior (e.g., lifecycle stage, needs, replenishment patterns, engagement history) and integrate them into generative and deep learning models to enable long-horizon reasoning.

- Experiment scalable representations of customer and household behavior that summarize long engagement history into compact states, supporting efficient, incremental inference in large-scale inference.

- Design and apply post-training optimization techniques (e.g., auxiliary objectives, preference modeling, offline reinforcement learning or policy optimization) to align model behavior with long-term engagement, satisfaction, and retention metrics.

- Develop robust evaluation frameworks combining offline metrics, counterfactual analysis, and online experimentation to measure both immediate impact and long-term customer outcomes.