RAG Document Chatbot

Upload any PDF or TXT and ask questions. Uses free local embeddings, FAISS vector search, and OpenAI GPT for answers with source citations.

---

Business Problem

Professionals waste hours scanning long documents. This tool lets anyone ask plain-English questions and get precise, cited answers in seconds — from contracts, reports, manuals, or research papers.

---

Architecture

Upload PDF/TXT → Chunk → Embed (free, local) → FAISS → Query → OpenAI GPT → Answer + Citations

---

Key Features

Feature	Detail
Free embeddings	all-MiniLM-L6-v2 via sentence-transformers — no API cost
Local vector search	FAISS in-memory — no external database
Conversational memory	Remembers last 5 turns
Source citations	Every answer links to source chunks
Model selector	Switch between gpt-3.5-turbo, gpt-4o-mini, gpt-4o

---

Use Cases

• Legal contracts — "What are the termination clauses?"
• Financial reports — "What was Q3 revenue growth?"
• Research papers — "What methodology did the authors use?"
• HR policies — "What is the leave policy?"

---

Quickstart

git clone https://github.com/deepanshu0110/rag-document-chatbot.git
cd rag-document-chatbot
python -m venv venv && source venv/bin/activate
pip install -r requirements.txt
cp .env.example .env  # add your OpenAI API key
streamlit run app.py

---

Tech Stack

Python · LangChain · FAISS · sentence-transformers · OpenAI · PyPDF · Streamlit

---

Roadmap

• DOCX, CSV, Markdown support
• Persistent FAISS index
• Ollama local LLM (fully offline)
• Streamlit Cloud deployment

---

Author

Deepanshu Garg — Freelance Data Scientist

• GitHub: @deepanshu0110
• Hire: freelancer.com/u/deepanshu0110

MIT License