This project enables the compilation of SQL-like code into executable workflows on a Kubernetes (K8s) cluster, achieving AI engine training/prediction results with nothing more than writing SQL. It supports databases such as MySQL, MariaDB, TiDB, Hive, and frameworks like TensorFlow, Keras.