README.md

Sentiment Analysis API

A production-ready REST API for analyzing text sentiment using transformer models, containerized with Docker.

Features

• Fast sentiment analysis using DistilBERT transformer model
• RESTful API with automatic interactive documentation
• Docker containerized for consistent deployment
• Health check endpoints for production monitoring
• Input validation with Pydantic models
• Sub-100ms inference after model warm-up

Tech Stack

• Backend: FastAPI (async Python web framework)
• ML Model: DistilBERT via HuggingFace Transformers
• Containerization: Docker
• API Documentation: Auto-generated with Swagger UI

Quick Start

Prerequisites

• Docker Desktop installed
• 4GB RAM minimum

Run with Docker

# Build image
docker build -t sentiment-api:v1.0 .

# Run container
docker run -d -p 8000:8000 --name sentiment-api sentiment-api:v1.0

# View logs
docker logs sentiment-api

Access API

• Interactive Docs: http://localhost:8000/docs
• Health Check: http://localhost:8000/health

🔌 API Usage

Example Request

curl -X POST "http://localhost:8000/analyze" \
  -H "Content-Type: application/json" \
  -d '{"text": "I love this product!"}'

Example Response

{
  "text": "I love this product!",
  "sentiment": "POSITIVE",
  "confidence": 0.9998,
  "processing_time_ms": 45
}

API Endpoints

Method	Endpoint	Description
GET	/	Health check with version info
POST	/analyze	Analyze text sentiment
GET	/health	Kubernetes-style health endpoint

Project Structure

sentiment-api/
├── src/
│   └── main.py          # FastAPI application
├── tests/               # Unit tests
├── Dockerfile           # Container definition
├── .dockerignore       # Docker build exclusions
├── requirements.txt     # Python dependencies
└── README.md

Development

Local Setup (without Docker)

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

# Install dependencies
pip install -r requirements.txt

# Run locally
uvicorn src.main:app --reload

Run Tests

pytest tests/ -v

Docker Details

• Base Image: python:3.11-slim
• Image Size: ~1.2GB (includes PyTorch + transformers)
• Health Check: Configured for production monitoring
• Multi-stage optimized for faster rebuilds

Performance

• First request: 30-60s (model download)
• Subsequent requests: < 100ms
• Memory usage: ~500MB (model in RAM)
• Concurrent requests: 10-20 (CPU-bound)

Future Enhancements

• GPU support for faster inference
• Model caching layer (Redis)
• Rate limiting
• Authentication
• Batch inference endpoint
• Multiple model support
• Kubernetes deployment configs

License

MIT License - feel free to use for your projects!

Author

Your Name

• GitHub: @simplyarfan
• LinkedIn: Syed Arfan Hussain

Acknowledgments

• HuggingFace for the DistilBERT model
• FastAPI for the excellent framework