Turbopuffer
Serverless vector search on object storage
turbopuffer is a serverless vector and full-text search database designed for fast and scalable search capabilities, particularly for AI applications and large data retrieval tasks. Founded by Simon Hørup Eskildsen and Justine Li, the company aims to make every byte searchable by connecting AI with extensive, fresh data. Based in Canada with remote operations and offices in Ottawa, Vancouver, and New York City, turbopuffer emphasizes cost efficiency and high performance.
The unified search platform combines vector and full-text search features, offering automatic scaling, low latency, and support for billions of vectors. It can handle significant production workloads, including over 1 trillion documents and high query throughput. Ideal use cases include AI applications, semantic search, recommendation systems, and large-scale document retrieval. The platform is available by referral or invitation.