How ayushigupta9723/swiggy-food-delivery-time-prediction works
Project type and purpose
• This is a full end-to-end Machine Learning plus MLOps project (“Swiggy Delivery Time Prediction”) that ingests raw data, processes it, trains multiple models, tracks experiments, registers a best model, packages the inference API in Docker, and deploys it on AWS EC2 via a CI/CD pipeline.
File structure
• Top-level:
• data/ (raw → processing with DVC)
• src/ (modular ML pipeline code: data_cleaning, data_preparation, features, train, evaluation, register_model)
• notebooks/ (EDA & experimentation)
• models/ (persisted artifacts)
• reports/ (metrics & figures)
• Scripts/ (utility scripts, e.g., promote_model_to_prod.py)
• app.py (Flask/FastAPI service for inference)
• dvc.yaml & dvc.lock (DVC pipeline stages)
• .github/workflows/ci_cd.yaml (GitHub Actions CI/CD)
• Dockerfile (container definition)
• requirements.txt, setup.py
• tests/ & test_environment.py (unit/integration tests)
README for additional insights
• It describes the DVC pipeline stages (Preprocess, Feature Engg, Train, Evaluate, Registry), MLflow + DagsHub tracking, GitHub Actions automated workflow (install, DVC pull, dvc repro, pytest, promote model, build & push Docker, deploy on EC2).
• It lists the end-to-end flow from data collection to deployment.