🧠 RagApp – End-to-End Retrieval-Augmented Generation (RAG) System

RagApp is a full-stack, extensible project for building Retrieval-Augmented Generation (RAG) systems — from data ingestion and vector indexing to LLM-based response generation and deployment.

---

---

This repo is designed for:

• Hands-on learning (understand each piece of a RAG system)
• Modular experimentation (swap databases, embeddings, and chunking logic)
• Scaling to production (async pipelines, orchestration, and deployment)

🔧 Version 2.0.0 is fully ready — complete migration to PostgreSQL + PGVector, dual vector DB support with Qdrant, custom chunking logic, and orchestration improvements.

---

✅ Highlights – Version 2.0.0

1. Full Migration to PostgreSQL + PGVector ✅

   • MongoDB fully replaced.
   • PGVector now fully implemented for vector search.

2. Dual Vector DB Support 🟢

   • Qdrant remains fully functional alongside PGVector.
   • Developers can switch between PGVector and Qdrant easily via a single .env variable (VECTOR_DB_PROVIDER).
   • Unified interface ensures seamless vector DB operations.

3. Custom Chunking Logic ✂️

   • LangChain is no longer required.
   • Custom chunking tailored to RagApp’s workflow for better performance and flexibility.
   • Developers can optionally enable LangChain chunking if desired.

4. Orchestration & Performance Improvements ⚡

   • Optimized async pipelines and indexing workflows.
   • Faster and more robust search and ingestion.

5. Deployment Ready 🚀

   • v2 backend is fully stable.
   • Next steps: integrate Celery + Redis for distributed task management and scaling.

---

📦 Tech Stack (v2)

• Backend: FastAPI + Uvicorn (async-first, OpenAPI-ready)
• Vector DB: PostgreSQL + PGVector (primary) + Qdrant (optional, fully supported)
• Embeddings / LLMs: OpenAI, Ollama, Cohere, Sentence Transformers
• Chunking: Custom logic (LangChain optional)
• Dockerized services: PostgreSQL, Qdrant
• Unified vector DB interface (switch PGVector ↔ Qdrant via .env)

---

⚡ Quickstart

1. Clone & Setup

git clone https://github.com/silvaxxx1/RagApp.git
cd RagApp

2. Install Dependencies

uv init
uv add -r requirements.txt

3. Configure Environment Variables

cp uv.example .env

• Set API keys for OpenAI, Ollama, etc.
• Choose your vector DB provider:

VECTOR_DB_PROVIDER=pgvector   # or "qdrant"

4. Run Services (Docker)

cd docker
docker-compose up -d

5. Run the Backend

uvicorn main:app --reload --host 0.0.0.0 --port 5000

Access Swagger UI → http://localhost:5000/docs

---

🗺️ Roadmap (v2.0.0)

• Full migration to PostgreSQL + PGVector
• Dual vector DB support (PGVector + Qdrant)
• Unified DB interface for seamless switching
• Custom chunking logic implemented (LangChain optional)
• Orchestration & async pipeline improvements
• Background tasks with Celery + Redis
• Advanced RAG strategies (re-ranking, hybrid, multi-query)
• Production deployment templates (Docker/K8s, CI/CD)

---

🤝 Contributing

Fork, clone, and build along! Ideas, PRs, and discussions are welcome as we evolve RagApp into a production-grade RAG template for the community.

---

📄 License

MIT License – see LICENSE

---

This README now clearly communicates:

• v2 is fully ready ✅
• PGVector fully implemented and MongoDB removed
• Dual vector DB support with easy switching via .env
• Custom chunking replaces LangChain but remains optional
• Orchestration and async pipelines improved

---