orbital

Orbital converts trained scikit-learn pipelines into SQL queries that can execute directly in a database without requiring Python. This enables deploying machine learning models where only database access is available.

The package automatically translates both preprocessing steps (like StandardScaler) and model predictions (including linear models, decision trees, random forests, and gradient boosting) into equivalent SQL. This solves the deployment problem of running scikit-learn models in environments where Python isn’t available or when you need to score data at scale directly in the database. The generated SQL produces identical predictions to the original scikit-learn pipeline.