🎭 Sentiment Classification - MLOps Pipeline

Production-grade sentiment analysis with automated experimentation, testing, and deployment

---

🎯 Overview

End-to-end MLOps system for sentiment analysis featuring automated experimentation, intelligent model promotion, and cloud deployment.

---

🌈 User Interface

---

🌈 Video Demo

▶️ Click to watch demo

---

🌈 Architecture and Workflow Diagrams

---

✨ Key Features

🔬 SYSTEMATIC EXPERIMENTATION

• Tested multiple models with BoW and TF-IDF
• Tracked all experiments using MLflow

🔄 AUTOMATED ML PIPELINE

• End-to-end DVC pipeline for data → model
• Fully reproducible with versioned parameters

🎯 SMART MODEL PROMOTION

• Automatically promotes only high-quality models
• Uses MLflow registry for staging and production

🚀 COMPLETE CI/CD PIPELINE

• Automated builds and deployments via GitHub Actions
• Dockerized deployment on AWS EC2

🧪 COMPREHENSIVE TESTING

• Validates model performance and API endpoints
• Prevents faulty models from being deployed

🌐 PRODUCTION-READY APPLICATION

• FastAPI app for real-time sentiment prediction
• Clean UI with health and confidence checks

---

🛠️ Tech Stack

• Machine Learning: Pandas & NumPy, NLTK
• Mlops Tools: MLflow, DVC, DagShub
• Deployement & CICD: Docker, GitHub Actions, AWS (EC2, ECR) , FastAPI

---

📁 Project Structure

Sentiment-Classification/
├── sentiment_classification/
│   ├── data/              # Data ingestion & preprocessing
│   ├── features/          # Feature engineering (BoW/TF-IDF)
│   ├── modeling/          # Training, evaluation, registry
│   └── connections/       # AWS S3 integration
├── fastapi_app/
│   ├── app.py            # FastAPI application
│   └── templates/        # Web interface
├── notebooks/            # Experimentation notebooks
├── scripts/
│   └── promote_model.py  # Smart model promotion
├── tests/                # Unit tests
├── data/                 # Dataset (tracked by DVC)
├── models/               # Saved models
├── .github/workflows/    # CI/CD pipeline
├── dvc.yaml              # DVC pipeline definition
└── Dockerfile            # Container configuration

---

🚀 Setup & Deployment

Want to run this project?

👉 Complete Setup Instructions

Includes local setup, DVC pipeline execution, MLflow tracking, Docker deployment, and AWS deployment guide.

---

🎓 What I Learned

• Building reproducible ML pipelines with DVC
• Experiment tracking and model versioning with MLflow
• Conditional deployment strategies
• CI/CD for ML systems
• Docker containerization best practices
• AWS cloud deployment (ECR + EC2)
• Writing production-ready ML code
• Comprehensive testing for ML systems

---

🔮 Future Enhancements

• Kubernetes: Migrate to K8s for auto-scaling
• Redis Caching: Cache predictions for faster responses
• Authentication: Add user management with OAuth2/JWT
• Monitoring: Implement Prometheus + Grafana dashboards
• A/B Testing: Compare model versions in production
• Explainability: Add SHAP/LIME for prediction explanations

---

👤 Author

Harsh Patel
📧 code.by.hp@gmail.com
🔗 GitHub • LinkedIn

---

⭐ Star this repo if you find it useful