The online models that mlrequest provide currently do not require explicit hyperparameter tuning to get the best performance. This is because mlrequest internally tunes hyperparameters so that you do not have to. As training data becomes available to your model, it will automatically adapt to the best hyperparameters to maximize performance.
Abstracting away hyperparameters allows you to focus on feature construction. Training the best model requires you to give it the best data. Construct features that communicate your data to your model in the most efficient way.