A data scientist employs MLflow for tracking their machine learning experiment. As part of each MLflow run, they conduct hyperparameter tuning. The scientist wishes to organize one parent run for the tuning procedure and have a child run for each unique combination of hyperparameter values. They manually initiate all parent and child runs using 'mlflow.start_run()'. Which methodology should the data scientist adopt to achieve this MLflow run organization? Choose only ONE best answer.