Ludwig v0.4.1 introduces an experimental AutoML module that integrates auto-config generation with hyperopt on Ray Tune, aiming to simplify the development and deployment of deep learning models for various data types, such as tabular, natural language processing, and computer vision. This open-source framework allows users to train models without writing code by specifying configurations in a YAML file, leveraging Horovod and Ray for scalability, and using MLflow for production deployment. The AutoML module focuses on tabular datasets, employing heuristics derived from extensive hyperparameter tuning experiments to produce models with competitive accuracy compared to expert-developed models, while reducing the computational resources and time needed. Ludwig AutoML positions itself as a more transparent "glass-box" solution compared to traditional AutoML systems, maintaining user control and flexibility in model creation and refinement. Future developments aim to extend its capabilities to image and text datasets and offer a managed platform for automated machine learning through Predibase.