VittoriaDB: Zero-config local vector database in a single Go binary (github.com)

VittoriaDB is a new zero-configuration, high-performance embedded vector database packed into a single Go binary, designed to simplify local AI development and production deployments. Addressing the complexity of cloud-based vector databases and the limitations of in-memory solutions, VittoriaDB offers out-of-the-box functionality with no dependencies, enabling rapid prototyping, edge deployment, and secure local vector management without cloud costs or complex setup. Technically, VittoriaDB leverages efficient HNSW (Hierarchical Navigable Small World) indexing for sub-millisecond similarity searches, supports multiple distance metrics (cosine, Euclidean, dot product, Manhattan), and provides ACID-compliant file-based persistence with write-ahead logging and crash recovery. Its dual interface—featuring a REST API and a native Python client that auto-manages the Go binary—makes it accessible for various workflows. Additional capabilities include metadata filtering, batch operations, multiple index types (exact flat and approximate HNSW), a built-in web dashboard, and planned seamless integration with embedding models for AI applications like retrieval-augmented generation, semantic search, and recommendation systems. Cross-platform support (Linux, macOS, Windows) and a rich set of CLI commands and documentation enhance developer experience, positioning VittoriaDB as a powerful, portable, and production-ready solution for vector similarity search in local and edge AI environments.