Applied Science II, Product Knowledge GenAI

The Catalog Services Product Knowledge team is seeking an Applied Scientist for the Catalog Services organization. Our vision is simple: build AI systems that are capable of a deep product understanding, so we can organize and scale the catalog metadata (schema) for Amazon e-commerce catalog worldwide. This is a complex problem because the magnitude of products entities (attributes, values, constraints) to be modeled to cover all the Amazon products worldwide. You will work on initiatives (models, artifacts) aim to solve the problem of producing Catalog schema with less reliance on humans and deliver them into the Amazon production ecosystem. Your efforts will build a robust ensemble of ML and GenAI techniques that will scale our catalog artifacts with a high precision across countries and languages. The scientist will own investments in machine learning, natural language processing, GenAI, to solve real world problems at scale. The team's output affects the velocity at which we build product schema and support the largest e-commerce catalog and impact million of customers. The team builds solutions ranging from automatic generation of product metadata, classification of entities, validation of concepts against customer traffic, creation of agents solving complex tasks mimicking human decisions at high precision, etc; all these developments drive true understanding of products at scale. The ideal candidate has deep expertise in one or several of the following fields: Generative AI, Agents, LLMs, Web search, Applied/Theoretical Machine Learning, Deep Neural Networks, Classification Systems, Clustering, Natural Language Processing. S/he has a strong publication record at relevant academic venues and proven experience in launching products/features in the industry. Key job responsibilities - Formulate open research problems at the intersection of GenAI, multimodal reasoning, and large-scale information retrieval—defining the scientific questions that transform ambiguous, real-world catalog challenges into models applied to production with high-impact - Push the boundaries of models and agentic architectures by designing novel approaches to catalog understanding, schema inference, where the problem complexity (billions of products) demands methods that don't yet exist - Make frontier models reliable—advancing uncertainty calibration, confidence estimation, and interpretability methods so that frontier-scale GenAI systems can be trusted for autonomous catalog decisions - Own the full research lifecycle from problem formulation through production deployment, designing rigorous experiments, iterating on ideas rapidly, and seeing your research directly improve data and catalog operations - Shape the team's research vision by defining technical roadmaps that balance foundational scientific inquiry with measurable product impact - Represent the team in the broader science community, publishing findings, delivering tech talks, and staying at the forefront of GenAI, and agentic system research About the team The team's mission is to infer knowledge, understand, and derive product schema for all Amazon products entering the Catalog. The work is critical to power drive policies on how products will be merchandised, guide Selling Partners, inform models how to infer attributes. All this information drives the navigational Taxonomy, Search and Detail Page experiences, impacting million of customers. The scientist collaborates closely with teams across the organization and outside the Catalog (Search, Personalization, etc) that rely on this team's developments.