Continual lets you build end-to-end ML applications inside your data warehouse. Users define feature sets using SQL, set up a regression or classification problem, and then let Continual predict the target column. Importantly, an analytics engineer can ship this end-to-end without ever needing to set up a Python runtime or any additional ML infrastructure.

Under the hood, Continual solves problems that stymie many ML projects. For example, Continual provides extremely reliable model runs and table updates. Many ML projects fail not because the classifier/regressor was bad, but because the supporting infrastructure (cough, Python script on a crontab, cough) breaks. Continual also implements best practice bookeeping (keeping immutable copies of feature sets, saving inference-time model predictions, etc.) from the start, something many teams don't realize they need until they really wish someone had implemented it months ago.

Based In
Scroll Down

The Challenge