QuotationManagementAPI

Python FastAPI backend for Quotation Management System with Vector Embeddings and RAG-based Natural Language Q&A. This system allows you to store quotation data and query it using natural language questions.

Features

• 🔍 Natural Language Q&A: Ask questions about quotations in plain English
• 🧠 Vector Embeddings: Uses sentence transformers for semantic search
• 🗄️ ChromaDB: Persistent vector database for fast retrieval
• 🚀 FastAPI: Modern, fast web framework with automatic API documentation
• 📊 Comprehensive Data Model: Supports customer, quotation, and item tracking
• 🔄 Bulk Operations: Add multiple quotations efficiently

Architecture

This system implements a Retrieval Augmented Generation (RAG) architecture:

1. Data Ingestion: Quotation data is converted to text and embedded using sentence-transformers
2. Vector Storage: Embeddings are stored in ChromaDB for efficient semantic search
3. Query Processing: Natural language questions are embedded and used to find similar quotations
4. Answer Generation: Relevant quotations are retrieved and formatted into natural language answers

Technology Stack

• FastAPI: REST API framework
• SQLAlchemy: Database ORM
• Sentence Transformers: Vector embeddings (model: all-MiniLM-L6-v2)
• ChromaDB: Vector database
• Pydantic: Data validation

Installation

# Clone the repository
git clone https://github.com/bsuraj23/QuotationManagementAPI.git
cd QuotationManagementAPI

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

Quick Start

1. Run the API server

python main.py

The API will be available at http://localhost:8000

2. Access API Documentation

Open your browser and go to:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

API Endpoints

1. Add Single Quotation

curl -X POST "http://localhost:8000/quotations/add" \
  -H "Content-Type: application/json" \
  -d '{
    "id": 1,
    "customername": "John Industries",
    "customerphone": "+91-9876543210",
    "customeremail": "john@industries.com",
    "quotationcode": "QT-2025-001",
    "quptationstatus": "pending",
    "quotationtotalamount": 15000.00,
    "itemname": "Bearing 6205",
    "itembrand": "SKF",
    "itemspecifications": "Deep groove ball bearing",
    "itemquantity": 50,
    "itemsellingprice": 300.00,
    "sellername": "Indispare"
  }'

2. Query Using Natural Language

# Simple GET query
curl "http://localhost:8000/query-simple?question=What quotations do we have for bearings?"

# POST query with more options
curl -X POST "http://localhost:8000/query" \
  -H "Content-Type: application/json" \
  -d '{
    "question": "Show me quotations for John Industries",
    "n_results": 5
  }'

3. Example Natural Language Questions

• "What quotations do we have for customer John?"
• "Show me all items with price above 10000"
• "Which quotations are pending?"
• "What items did we quote for bearings?"
• "Show me quotations from seller Indispare"
• "What are the most expensive items?"
• "Which customers have the most quotations?"

4. Bulk Add Quotations

curl -X POST "http://localhost:8000/quotations/bulk-add" \
  -H "Content-Type: application/json" \
  -d '[{
    "id": 1,
    "customername": "ABC Corp",
    "itemname": "Bearing",
    "itemsellingprice": 500
  },
  {
    "id": 2,
    "customername": "XYZ Ltd",
    "itemname": "Gear",
    "itemsellingprice": 750
  }]'

5. Get Statistics

curl "http://localhost:8000/stats"

Database Schema

The system uses the following fields from your quotation_items table:

• Customer Info: customername, customerphone, customeremail, customerid, customercode
• Quotation Info: quotationid, quotationcode, quptationstatus, quotationtotalamount, quotationtermsconditions
• Item Info: itemname, itemspecifications, itembrand, itemquantity, itemdeliverydate
• Pricing: itemlistingprice, itemsellerdiscount, itemcustomerdiscount, itempurchaseprice, itemsellingprice
• Additional: itemproductid, itemhsncode, itemuom, itemtaxpercent
• Seller Info: sellername, sellerphone

How Vector Embeddings Work

1. Text Creation: Each quotation item is converted to a searchable text format:

   Customer: John Industries | Email: john@industries.com | Item: Bearing 6205 | Brand: SKF | Selling Price: 300.00
   

2. Embedding Generation: The text is converted to a 384-dimensional vector using sentence-transformers

3. Storage: Vectors are stored in ChromaDB with metadata

4. Query: Your question is also converted to a vector and compared with stored vectors using cosine similarity

5. Retrieval: The most similar quotations are retrieved and formatted into an answer

Use Cases

• Sales Teams: Quickly find quotations by customer, product, or status
• Customer Support: Answer customer queries about their quotations
• Analytics: Discover patterns in quotation data
• Reporting: Generate insights from historical quotation data

Project Structure

QuotationManagementAPI/
├── app/
│   ├── models.py              # SQLAlchemy database models
│   └── embedding_service.py   # Vector embedding and RAG service
├── main.py                    # FastAPI application
├── requirements.txt           # Python dependencies
├── .gitignore
└── README.md

Contributing

Feel free to open issues or submit pull requests!

License

MIT License