Amgix

Amgix is an open-source hybrid search system that handles ingestion, embedding, and hybrid retrieval behind a single REST API. It replaces the need to combine queues, a vector database, and ranking or fusion logic in an application, handling queueing, deduplication, distributed locking, retries, and embeddings internally.

Features include server-side fusion of dense vectors, sparse models such as SPLADE, and keyword tokens on one collection; WMTR (Weighted Multilevel Token Representation) for lexical views including surface form, normalized tokens, and character-level patterns; an Amgix Dashboard with metrics endpoints exposing JSON and Prometheus metrics; and encoder workers that self-orchestrate by capacity and demand. Storage can run on PostgreSQL or MariaDB, with Qdrant supported for larger scale while keeping the API unchanged. Amgix One bundles the system into a single container, and deployments can scale into API, ingestion, query, communication, and storage tiers.

Amgix is aimed at developers integrating search, operators running deployments, and end users querying real-world data. It is distributed as open-source software, with Amgix One available as a single-container package.