📌 Project Overview

HabiMate 🌿

An AI-powered RAG (Retrieval-Augmented Generation) chatbot built with Streamlit and LangChain, designed to help users make informed, data-driven decisions for safer homes and resilient communities.

---

Overview

HabiMate uses a FAISS vector store to retrieve relevant context from local PDF documents and passes it to Google Gemini (gemini-2.5-flash-lite) to generate accurate, grounded answers. It features a clean chat UI built entirely in Streamlit.

---

Tech Stack

Layer	Technology
Frontend / UI	Streamlit
LLM	Google Gemini 2.5 Flash Lite (via LangChain)
Embeddings	sentence-transformers/all-MiniLM-L6-v2 (HuggingFace)
Vector Store	FAISS
RAG Framework	LangChain
Backend (API)	FastAPI
Data Source	PDF documents

---

Project Structure

├── .devcontainer/          # Dev container config
├── backend/                # FastAPI backend
├── data/
│   └── pdfs/
│       └── family_database_ovg_vi_dec_2024.pdf
├── vectorstore/            # FAISS index (generated)
├── app.py                  # Main Streamlit app
├── requirements.txt
├── .gitignore
└── README.md

---

How It Works

1. PDF documents are loaded from data/pdfs/ and chunked into segments
2. Each chunk is embedded using all-MiniLM-L6-v2 and stored in a FAISS index
3. When a user asks a question, the top 5 most relevant chunks are retrieved
4. The retrieved context + question are passed to Gemini via a LangChain RAG chain
5. The answer is displayed in the chat UI

---

Getting Started

1. Clone the repository

git clone https://github.com/your-username/habimate.git
cd habimate

2. Install dependencies

pip install -r requirements.txt

3. Set up your API key

Create a .env file in the root directory:

GOOGLE_API_KEY=your_google_api_key_here

4. Build the vector store

Make sure your PDFs are in data/pdfs/, then run your ingestion script to generate the FAISS index in vectorstore/faiss_index.

5. Run the app

streamlit run app.py

---

Backend API (FastAPI)

A lightweight FastAPI backend is also available for programmatic access:

uvicorn backend.main:app --reload

Endpoint:

GET /search?query=your+question+here

Returns:

{
  "result": "AI-generated answer based on retrieved context"
}

---

Environment Variables

Variable	Description
GOOGLE_API_KEY	Your Google Gemini API key

---

Requirements

Key dependencies (see requirements.txt for full list):

• streamlit
• langchain
• langchain-google-genai
• langchain-community
• faiss-cpu
• sentence-transformers
• python-dotenv
• fastapi
• uvicorn

---

Notes

• The chatbot answers only from the provided PDF context — it will not hallucinate answers outside the retrieved documents.
• The FAISS index must be built before running the app. The vectorstore/ folder is not committed to the repo (see .gitignore).