MLflow for Out-of-Sample prediction

Hey all,

Wanted to ask if anyone has been able to successfully combined something like the model-builder class with mlfow to perform out of sample prediction by importing a model in mflow

I have been able to use mlflow to save artifacts(inference frames, metrics, etc etc)-however, If possible I would like to save a custom tree/GP/etc model as an mlflow object if possible.

Thanks!