🤖 SQL Query Buddy

A conversational AI agent that converts natural language questions into SQL queries, executes them, and provides AI-driven insights.

Built with LangChain, FAISS VectorDB, and Gradio for the Codecademy GenAI & Agents Bootcamp Contest.

---

🎯 Features

• Natural Language to SQL - Ask questions in plain English, get accurate SQL queries
• RAG-Powered - Uses vector database to semantically search table schemas
• SQL Explanations - Beginner-friendly explanations of what each query does
• AI Insights - Analyzes query results and provides business insights
• Conversation Memory - Remembers context for follow-up questions
• Clean Chat Interface - Interactive Gradio UI for easy interaction

---

🛠️ Tech Stack

Component	Technology
AI Framework	LangChain
LLM	OpenAI GPT-3.5
Vector Database	FAISS
Embeddings	OpenAI Embeddings
Database	SQLite
Frontend	Gradio
Language	Python 3.13

---

📦 Installation

1. Clone the repository

git clone https://github.com/yourusername/sql-query-buddy.git
cd sql-query-buddy

2. Create virtual environment

python -m venv .venv
source .venv/bin/activate  # Mac/Linux
# OR
.venv\Scripts\activate     # Windows

3. Install dependencies

pip install langchain langchain-openai langchain-community python-dotenv sqlalchemy faiss-cpu gradio

4. Set up environment variables

Create a .env file in the project root:

OPENAI_API_KEY=your_openai_api_key_here

5. Initialize the database

cd database
python setup_database.py
cd ..

---

🚀 Usage

Run the application

python app.py

This will launch the Gradio interface in your browser.

Example Questions

• "Show me all customers from California"
• "What are the top 3 customers by total spending?"
• "Which product category generates the most revenue?"
• "Show total sales per region for 2025"
• "How many unique products were sold in January?"
• "Show the trend of monthly revenue over time"
• "Find the average order value for returning customers"

---

🗄️ Database Schema

The project uses a retail commerce database with 4 tables:

customers - Customer information (name, email, region, signup date)

products - Product catalog (name, category, price)

orders - Order records (customer, date, total amount)

order_items - Line items (links orders to products with quantity)

---

🧠 How It Works

1. User asks a question in natural language

2. Vector search finds relevant table schemas from FAISS database

3. LLM generates SQL using the retrieved schema context

4. Query executes against the SQLite database

5. AI analyzes results and provides insights

6. Conversation saved to memory for follow-up questions

---

🔑 Key Components

sql_agent.py

• create_schema_documents() - Creates searchable documents for each table
• create_vector_store() - Embeds schemas in FAISS vector database
• get_relevant_schemas() - Retrieves relevant tables based on question
• ask_question() - Main RAG pipeline: retrieve → generate → execute → insight

app.py

• Gradio chat interface
• Integrates with sql_agent functions
• Displays SQL, results, and insights

---

👤 Author

Rubin - Software Engineer Codecademy GenAI & Agents Bootcamp Participant

---

🏷️ Tags

#CodecademyGenAIBootcamp #RAG #LangChain #VectorDB #SQL #AI #Gradio