Billion-Scale Vector Search on Object Storage

This 34-minute conference talk from Data Council features Simon Hørup Eskildsen and Mickey Liu sharing their expertise on powering large-scale vector search systems using object storage. Discover the database architecture and pipeline engineering that enables Notion's semantic search functionality across billions of vectors. Learn valuable insights from the collaboration between turbopuffer and Notion on managing massive vector workloads and implementing object-storage native indexing—critical knowledge for teams connecting Large Language Models to data or building Retrieval-Augmented Generation systems at scale. The presentation is part of Data Council's Day 1 series focused on Data Engineering and Infrastructure.