A real-time, production-ready Retrieval-Augmented Generation (RAG) system built using open-source LLMs, vector databases, and FastAPI.
This project follows industry-standard architecture and is suitable for:
- Enterprise document search
- Internal knowledge bots
- Policy & report Q&A
- Learning modern LLM systems
- Open-source LLM (Ollama + Mistral / LLaMA)
- Vector search using Qdrant
- Multi-format document ingestion:
- PDF, CSV, TXT, JSON, DOCX, HTML, MD, PPTX
- Async background ingestion
- Sentence-aware chunking
- Normalized embeddings
- Production-ready API
- Dockerized deployment
User │ │ (upload / ask) ▼ FastAPI │ ├── Ingestion Pipeline │ ├── Loader │ ├── Chunker │ └── Embedder │ ├── Vector DB (Qdrant) │ └── RAG Pipeline ├── Retrieval └── LLM (Ollama)