A compiler that extends SQL with AI capabilities to train, predict, and evaluate machine learning models directly from SQL statements.
SQLFlow is a compiler that extends SQL with machine learning capabilities, allowing users to train, predict, evaluate, and explain AI models directly from SQL statements. It solves the fragmentation between data processing and AI development by enabling SQL-skilled engineers to build ML applications without switching to languages like Python or R.
Data engineers, data scientists, business analysts, and developers who use SQL for data management and want to integrate machine learning workflows without learning new programming languages.
Developers choose SQLFlow for its ability to unify SQL and AI in a single, extensible framework that supports multiple databases and ML toolkits, simplifying the ML lifecycle and reducing engineering overhead compared to proprietary SQL engines or disjointed toolchains.
Brings SQL and AI together.
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Extends SQL with AI job syntax for training, prediction, and evaluation, enabling elegant ML code like SELECT ... TO TRAIN DNNClassifier without switching languages.
Supports multiple SQL engines including MySQL, Hive, and TiDB, allowing integration with diverse data sources as listed in the README.
Integrates with TensorFlow, Keras, and XGBoost, providing access to state-of-the-art machine learning toolkits for various models.
Compiles SQL programs into Argo workflows for distributed execution on Kubernetes, facilitating scalable and production-ready ML deployments.
Requires a Kubernetes cluster and Argo setup, adding operational complexity and making it unsuitable for environments without container orchestration.
Supports specific ML toolkits; expanding to new frameworks relies on community contributions, as noted in the roadmap, which can slow adoption for niche use cases.
Building and deploying involves multiple steps detailed in the Build and Quick Start documentation, creating a barrier to entry for quick experimentation.