All open source
Open SourceRAG6k

Vespa

Big-data serving engine fusing vector, text, and structured search.

Vespa is an open-source platform from Yahoo for storing, searching, and serving large datasets with combined vector, lexical, and structured filtering plus in-engine ranking and ML inference. It runs hybrid retrieval and re-ranking at scale in a single system, with first-class support for approximate nearest-neighbor search alongside BM25 and tensor computations. It is notable as a production-grade engine for large RAG and recommendation workloads that need more than a pure vector store.

Repository

vespa-engine/vespa

Language

Java

What you'd build with it

  • Serving hybrid (vector plus keyword plus filter) retrieval with custom ranking for large-scale RAG systems
  • Running first-stage retrieval and learned re-ranking together inside one engine to cut serving complexity
  • Powering recommendation and search at scale where vectors must combine with structured and text signals

Tags

vector-searchhybrid-searchrankingsearch-engine