Sample model: download from external MLFLow

Download a model tracked in an external MLFlow. This is currently the only way to get a model from an external model registry into Azure deployment.

import mlflow
import joblib, os

If needed, set AWS credentials in local environment to access MLFlow on AWS

#os.environ["AWS_ACCESS_KEY_ID"] = ""
#os.environ["AWS_SECRET_ACCESS_KEY"] = ""

mlflow.set_tracking_uri("http://andre-mlflo-a3oi3g4g7sh1-6b98eef0c99b7004.elb.eu-central-1.amazonaws.com")
model = mlflow.sklearn.load_model(f"models:/diabetes/1")

joblib.dump(model, "model.pkl")