MLOps Project

This repository contains an end-to-end example for deploying, tracking, and monitoring machine learning models. It demonstrates data preparation, model training, experiment tracking with MLflow, simple deployment options (FastAPI, Streamlit, Dash), and retraining orchestration. Certificate Key Concepts

• Model experimentation and tracking with MLflow
• Lightweight model serving with FastAPI
• Interactive apps with Streamlit and Dash
• Retraining and workflow automation examples (Airflow-ready script)

Repository Layout

• datasets/ — example CSV datasets used for training and augmentation
• Manage/ — training, retraining, augmentation, and MLflow tracking utilities
• ML/ — model training scripts (examples using RandomForest / other estimators)
• dep/ — small deployment/demo app scripts (FastAPI, Streamlit, Dash)
• mlruns/ — MLflow experiment runs and saved model artifacts

MLflow tracking & models

• Experiment runs are stored under Manage/mlruns/ (or top-level mlruns/ depending on how MLflow is configured).
• Saved models and environment files appear in Manage/mlruns/*/models/*/artifacts/.

Retraining / Automation

• Manage/retrain.py and Manage/retrainWithAirFlow.py contain example retraining logic. They can be adapted to run on a schedule (Airflow, cron) and to register new model versions with MLflow.

Datasets

• datasets/augmented_earthquake_data.csv — augmented earthquake dataset used in examples
• datasets/earthquake_alert_balanced_dataset.csv — balanced dataset for classification experiments
• datasets/flight_dataset.csv — flight-related example dataset