Snowflake is simplifying the development of ML models and full-stack apps for developers using Snowflake’s Data Cloud. This will be made possible partly through enhancements to the Python programming language in Snowpark.
Snowflake believes companies are placing a higher value on data with the advent of AI. “Snowflake makes it easier for developers to put that data to work so they can build powerful end-to-end machine learning models and full-stack apps native to the Data Cloud,” said Prasanna Krishnan, senior director of product management at Snowflake.
According to the data specialist, developers want to develop and deploy machine learning models within Snowpark. Therefore, the company is making improvements regarding the Python programming language. The focus on Python will result from the recent acquisition of Ponder.
Improvements in Snowpark
In private preview, Snowflake Notebooks is currently already visible. That brings a new and interactive interface to Snowpark. However, the real productivity improvement is in the notebooks available to developers and allowing developers to write, execute and train code.
Snowpark ML Modeling API will become generally available soon. Developers can work with the API to scale up feature engineering and simplify model training. To streamline work, there is the ability to link popular AI and ML frameworks.
Furthermore, there are improvements to the Snowpark ML Operations Enhancements that Snowflake will soon make generally available. First, the Snowpark Model Registry is thus, from now on, built on a native Snowflake model. Users benefit by making models in Snowflake scalable, securely deployable and manageable. Among the models will be deep learning and open-source language models from Hugging Face. Second, developers get access to an integrated Snowflake Feature Store in which ML features for model training are available.
Improvements in the app lifecycle
The entire app lifecycle is being taken in hand with the release of the Snowflake Native App Framework. The service will soon become generally available on AWS, while a public preview is planned for Azure. The service promises to provide all the components for app development, distribution, operationalization and monetization.
2 days ago
