RAGVerse_AI — Production-Grade Technical RAG Framework

Enterprise Retrieval-Augmented Generation (RAG) system specialized in technical documentation (LangChain, LangGraph, LangSmith, Langfuse, and LangFlow). Built with modular state graphs, multi-query expansion, hybrid sparse/dense retrieval, cross-encoder re-ranking, and persistent multi-turn session checkpoints.

---

🚀 Features

• Multi-Format Technical Ingestion: Parses .pdf, .md, .docx, .html, and .txt files into clean, metadata-enriched chunks.
• Advanced Retrieval Pipeline: Combines Multi-Query LLM expansion, Dense Similarity Search, BM25 Keyword Hybrid Search, and Cross-Encoder Re-Ranking (ms-marco-MiniLM-L-6-v2).
• Stateful Memory via LangGraph: Persistent multi-turn conversation memory backed by PostgreSQL checkpointers.
• Strict Grounding Rules: System prompts engineered to prevent hallucinations and enforce source attribution.
• Structured Application Logging: Rotating log files (logs/app.log) and telemetry hooks for LangSmith/Langfuse observability.

---

🛠️ Environment Setup

Copy .env.example to .env:

cp .env.example .env

Set the required environment variables:

GOOGLE_API_KEY=your_google_api_key
DATABASE_URL=postgresql://postgres:password@localhost:5432/rag_chatbot

# Embedding Configuration (Separate keys to avoid collisions)
GEMINI_EMBEDDING_MODEL_NAME=gemini-embedding-001
HUGGINGFACE_EMBEDDING_MODEL_NAME=BAAI/bge-base-en-v1.5

# Observability (LangSmith)
LANGCHAIN_TRACING_V2=true
LANGCHAIN_API_KEY=your_langsmith_key
LANGCHAIN_PROJECT=RAGVerse_AI

---

📂 Data Ingestion

Place your raw documentation files (.md, .pdf, .html, .docx) inside data/raw/documents/, then run:

python -m app.retrieval.ingest

---

💻 Running the Chatbot

Start the conversational CLI:

python run.py

---

🧪 Running Unit Tests

Execute the automated test suite:

pytest