SQLFlow is an open-source compiler that brings SQL and artificial intelligence together. It extends the SQL language with a compact syntax for training and using machine learning models, so data teams can build AI workflows without leaving the database world they already know. The project lives on GitHub with more than 5,100 stars and is developed by the SQLFlow community under the Apache-2.0 license.
In a typical organization, the people who understand the data speak SQL, while machine learning happens in Python notebooks far away. SQLFlow closes that gap: a statement like SELECT … TO TRAIN creates a model, and TO PREDICT applies it, with the compiler translating the extended SQL into a workflow that runs on Kubernetes. Under the hood it can drive TensorFlow, XGBoost and other engines, and it connects to data sources including MySQL, Hive, MaxCompute and SQLite. The goal is to make end-to-end machine learning, from feature extraction to model deployment, accessible through a single declarative language.
The project also provides a workflow-definition layer and an online playground for trying extended SQL in the browser, and it covers common algorithms out of the box, so a team can go from a SQL statement to a trained, serving model without writing separate orchestration code.
SQLFlow is free and open source under the Apache-2.0 license. There are no paid tiers or hosted upsells; you deploy it on your own infrastructure and use it without restriction.
SQLFlow is aimed at data engineers, analysts and database-centric teams that want to apply machine learning to warehouse data using familiar SQL instead of building separate Python pipelines.