Vespa is an open-source big data serving engine that enables low-latency computation over large structured, text and vector datasets at user-serving time. It provides storage, retrieval, ranking and real-time computation so applications can perform relevance ranking, personalization, recommendations and real-time decisioning at scale. Vespa can be self-hosted under an Apache 2.0 license or consumed as a serverless managed service (Vespa Cloud); it also offers SDKs and APIs (including pyvespa) for deployment, prototyping and integration with ML/embedding workflows. The platform is optimized for production-grade performance, tight relevance control and streaming retrieval patterns used in retrieval-augmented generation (RAG) and large-scale search applications.