Kubeflow seems pretty dead nowadays compared to tools like MLflow. Has anybody actually deployed it to production and can comment if it's been worth it?
Kind of, we started down the path of deploying with it, but realized that the number of moving pieces and chatter was too much. Ultimately decided to go for a managed solution. Databricks has a managed mlflow, sagemaker has an equivalent.
How does the price of the managed solutions scale for you?