How rajinikanthvadla-ai/mlops-lab-02-aws-sagemaker-end-to-end works
This repository implements a wine-quality SageMaker training and prediction workflow. Training can run either as an MLflow-enabled SageMaker pipeline or through a separate built-in XGBoost train-and-deploy script; a deployment script can also package the latest completed training artifact and create the prediction endpoint. The FastAPI application serves a form and forwards submitted features to that endpoint. The graph reflects the sampled source and README; the template’s behavior and deployment workflow details beyond the sampled deployment script are not independently verified.
Main entry points: train_with_mlflow.py MLflow trainer · create_pipeline.py Pipeline creator · deploy_latest_model.py Latest model deployer · app.py Prediction API
Components 10
Model training
- Pipeline creator
scripts/create_pipeline.py - MLflow trainer
scripts/train_with_mlflow.py - Direct train runner
scripts/run_train_deploy.py
Endpoint deployment
- Latest model deployer
deploy_latest_model.py
Prediction serving
- Prediction API
ml-api/app.py - Prediction form
ml-api/templates/index.html
AWS and tracking services
- S3 data and artifacts
- SageMaker
- MLflow tracking
Other components
- Wine user
Connections 14
- Wine user → Prediction API: submits features
ml-api/app.py - Prediction API → Prediction form: renders form
ml-api/app.py - Prediction API → SageMaker: invokes endpoint
ml-api/app.py - Pipeline creator → SageMaker: starts pipeline
scripts/create_pipeline.py - Pipeline creator → S3 data and artifacts: configures training input
scripts/create_pipeline.py - Pipeline creator → MLflow trainer: sets training entrypoint
scripts/create_pipeline.py - MLflow trainer → S3 data and artifacts: reads training channel
scripts/train_with_mlflow.py - MLflow trainer → MLflow tracking: logs runs
scripts/train_with_mlflow.py - MLflow trainer → SageMaker: writes model artifact
scripts/train_with_mlflow.py - Direct train runner → S3 data and artifacts: converts dataset
scripts/run_train_deploy.py - Direct train runner → SageMaker: trains and deploys
scripts/run_train_deploy.py - Latest model deployer → SageMaker: finds training job
deploy_latest_model.py - Latest model deployer → S3 data and artifacts: repackages artifact
deploy_latest_model.py - Latest model deployer → SageMaker: creates endpoint
deploy_latest_model.py