Principal Applied Scientist, Items and Relationships Platform
Core
Setting the scientific vision for product identity and relationships using multimodal large language models and generative AI to establish a trustworthy global product catalog.
Role type
Principal Applied Scientist (Entity Resolution & Relationship Inference)
Builds
Canonical product identity, matching, and relationship systems for Amazon's universal catalog
Domain
E-commerce, Generative AI, Multimodal Learning, Entity Resolution
Deliverable
production ML models
Required skills
Predictive modeling, Representation learning, Algorithm design, Calibration science, LLM evaluation, Java/C++/Python programming, Mentoring scientists
Preferred skills
Novel algorithm creation, State-of-the-art research, Peer-reviewed publications, Complexity analysis
Technologies
Multimodal LLMs, Generative AI, Java, C++, Python
Responsibilities
Define science direction for product identity and relationships, Build representations for sameness and typed relatedness, Own structural relationship science across variations and bundles, Define calibration and evaluation metrics, Partner with engineers on cost-aware modeling, Mentor senior and principal-track scientists
Seniority
Principal, strategy & mentorship