vespa-engine/vespa
🔧 Toolvespa-engine
AI-powered search platform for real-time vector and text search at scale.
Vespa is a highly scalable, real-time serving engine for search and recommendation systems. Built in Java, it supports both text and vector search with approximate nearest neighbor (ANN) indexing, allowing hybrid queries that blend lexical and semantic matching. Vespa can evaluate machine learning models (e.g., ONNX, TensorFlow) at query time, enabling personalized ranking and retrieval augmented generation (RAG). It features a distributed architecture with automatic data partitioning, replication, and failover. Advanced capabilities include tensors for multi-dimensional data, group evaluation, and SRE-friendly operations via REST APIs. Vespa is used in production at companies like Yahoo, serving billions of documents with sub-second latency.
💡Highlights
- ├─Unified text and vector search
- ├─Real-time ML model evaluation
- └─Scalable to billions of documents
🎯For
- ├─Search engineers
- ├─AI/ML engineers
- └─Data platform teams