Amgix

The product is an open-source hybrid search system that handles document ingestion, embedding, and hybrid retrieval behind a single REST API, removing the need to stitch together queues, vector databases, and fusion logic. It is for software developers and operators in B2B and data teams who build and run search on real-world enterprise data. It replaces glue code with server-side fusion of dense vectors, sparse models and keyword tokens and scales from a single container to distributed tiers.

Key features

• Ingestion, embedding and hybrid retrieval via REST API
• Queueing, deduplication, distributed locking and retries
• Server-side fusion of dense, sparse and keyword search
• SPLADE sparse models and keyword tokens
• WMTR weighted multilevel token representation
• Real-time dashboard and metrics endpoints
• JSON and Prometheus metrics scraping
• Single container to distributed tier scaling
• Autonomous MLOps encoder self-orchestration
• PostgreSQL and MariaDB vector storage support
• Qdrant integration for maximum scale
• Fused ranking with typeahead-level latency