ORAG: A Go-Native RAG Framework Combining Hybrid Retrieval and Document Ingestion

ORAG is an open-source, Go-native Retrieval-Augmented Generation (RAG) service framework designed to handle end-to-end workflows including document ingestion, hybrid retrieval, answer generation, and evaluation. The framework uses Hertz for its HTTP API, with PostgreSQL and Qdrant serving as the default storage and retrieval backends respectively. Ingestion is supported via JSON text import and multipart file upload, with jobs trackable through dedicated API routes such as /documents:import and /ingestion-jobs/{id}. Hybrid retrieval combines dense vector search via Qdrant and sparse full-text search via PostgreSQL, followed by fusion and reranking steps. Developers can explore the project locally using Docker Desktop and the make demo command, which spins up all dependencies and exercises the core ingestion and retrieval pipeline.