automl-gs provides a zero-code interface for building optimized ML/DL models for tabular prediction tasks. Users supply a CSV file and specify a target column; the tool handles data preprocessing, model architecture selection, and hyperparameter search across TensorFlow/Keras and XGBoost frameworks. The output is a fully native Python pipeline script with no platform lock-in, allowing integration into any prediction workflow. It supports TPU training via Google Colaboratory for faster model exploration.