HEKTOR is a high-performance vector search engine written in C++ with Python bindings. It combines BM25 full-text search with semantic/embedding-based retrieval for true hybrid search — built for AI agents, RAG pipelines, and production-grade retrieval infrastructure.

Key Features

• Hybrid Search: BM25 + dense vector retrieval in a single query
• C++ Core: Maximum throughput with minimal latency
• Python Bindings: Drop-in for AI/ML pipelines
• AI Infrastructure: Built for embedding-based retrieval at scale
• Zero Dependencies: Self-contained, easy to deploy