Demo Steps.md

Demo Steps

Servers

• mlflow: http://localhost:5000/
• airflow: http://localhost:8080/ (user-airflow; password-airflow)
• notebook: http://localhost:8888/ (password-notebook)
• serving: http://localhost:3000/

Preq

1. no registered model in mlflow
2. run ./clean.sh in examples/iris/serving/ folder

Development Env Demo

1. make up-experiment: boost experiment dev environment
2. open jupyter lab (find token in notebook docker)
3. walk through sandbox_experiment notebook
4. after done, make down to shutdown environment

Prod Serving Demo

1. make up-prod: boost prod demo environment
2. open Mlflow to check no registered model
3. open iris monitoring process to see the log
4. open airflow webserver: walk through iris dags
5. check training done, open iris_serving model_ver page to show loaded model version
6. open jupyter lab (find token in notebook docker)
7. open config_prod.yaml to show data_monitoring and performance_monitoring settings
8. use prod_demo notebook to hack production model