It's mostly for educational purposes, but why not make it bigger later on?
Repo: https://github.com/mburaksayici/RAG-Boilerplate
- Includes propositional + semantic and recursive overlap chunking, hybrid search on Qdrant (BM25 + dense), and optional LLM reranking.
- Uses E5 embeddings as the default model for vector representations.
- Has a query-enhancer agent built with CrewAI and a Celery-based ingestion flow for document processing.
- Uses Redis (hot) + MongoDB (cold) for session handling and restoration.
- Runs on FastAPI with a small Gradio UI to test retrieval and chat with the data.
- Stack: FastAPI, Qdrant, Redis, MongoDB, Celery, CrewAI, Gradio, HuggingFace models, OpenAI.
Blog : https://mburaksayici.com/blog/2025/11/13/a-rag-boilerplate.h...
I open-sourced the RAG boilerplate I’ve been using for my own experiments with extensive docs on system design.
And I have bunch of LLM+RAG blogs I post frequently last 2 months : https://mburaksayici.com/blog
It's mostly for educational purposes, but why not make it bigger later on? Repo: https://github.com/mburaksayici/RAG-Boilerplate - Includes propositional + semantic and recursive overlap chunking, hybrid search on Qdrant (BM25 + dense), and optional LLM reranking. - Uses E5 embeddings as the default model for vector representations. - Has a query-enhancer agent built with CrewAI and a Celery-based ingestion flow for document processing. - Uses Redis (hot) + MongoDB (cold) for session handling and restoration. - Runs on FastAPI with a small Gradio UI to test retrieval and chat with the data. - Stack: FastAPI, Qdrant, Redis, MongoDB, Celery, CrewAI, Gradio, HuggingFace models, OpenAI. Blog : https://mburaksayici.com/blog/2025/11/13/a-rag-boilerplate.h...