rag_api and rag-forge
These are complements: the first provides a production RAG API implementation while the second provides systematic benchmarking to optimize the chunking, embedding, and retrieval parameters that such an API would use.
About rag_api
danny-avila/rag_api
ID-based RAG FastAPI: Integration with Langchain and PostgreSQL/pgvector
Organizes embeddings by `file_id` to enable targeted, file-level vector retrieval with metadata filtering—particularly useful for multi-document RAG scenarios. Supports multiple embedding providers (OpenAI, Azure, Hugging Face, Bedrock, Ollama, Google) and vector backends beyond pgvector, with configurable chunking, batching, and async processing for scalability. Designed as a pluggable service for LibreChat but works as a standalone ID-based document indexing API with optional JWT authentication.
About rag-forge
Dyinu/rag-forge
Benchmark multiple chunking, embedding, and retrieval combinations for RAG pipelines to find the most effective setup without manual testing.
Related comparisons
Scores updated daily from GitHub, PyPI, and npm data. How scores work