Vector Store: Uber’s Embedding Platform - Feature Store Summit 2025

Vector Store: Uber’s Embedding Platform With the rise of Generative AI and large-scale semantic search, embeddings have become a foundational building block for enabling high-value ML and AI use cases across Uber. From personalized recommendations on Uber Eats to conversational assistants, embeddings now power Retrieval-Augmented Generation (RAG), our GenAI platform, semantic search, and predictive models at global scale. Over the past year, the Michelangelo team has elevated embeddings to first-class citizens within Uber's ML ecosystem—building a unified platform to simplify their generation, ingestion, versioning, and use across diverse applications. Today, this platform powers GenAI use cases and semantic search within the Uber App as well as internal systems. This talk introduces Vector Store, Uber’s scalable platform for managing the full lifecycle of embeddings—including offline/streaming generation, batch/real-time ingestion, standardized retrieval APIs, and automated model switching—all backed by centralized metadata and governance. Key Takeaways: - Evolution of Embedding Infrastructure: How Uber scaled from isolated RAG use cases to 150+ high-impact embedding-powered applications. - Operational Challenges: Lessons from solving real-world issues such as model versioning gaps, discoverability, access control, and ingestion complexity. - Platform Architecture: A deep dive into embedding generation via Hive/Kafka, centralized metadata APIs, and ingestion into OpenSearch and Lucene Plus. - Lifecycle Management: Managing embedding freshness, model switching, and standardized APIs for retrieval across use cases. About the Speakers: Divya Nagar is a Software Engineer on the Feature Engine team at Uber, where she leads the development of the Vector Store—the data plane powering embedding use cases across both Generative AI and CoreML applications. Her work focuses on designing and building scalable, reliable, and high-performance architecture for feature and vector stores. Divya is passionate about creating seamless, production-ready solutions that accelerate machine learning and AI development for engineers across Uber. Xiyuan is a software engineer at Uber’s Feature Store team, based in Sunnyvale, CA. With around three years of experience, Xiyuan has worked across various areas including feature ETL pipelines, feature transformations, RAG, vector stores, etc.