Snowflake unveils enhanced tech to simplify machine learning model development
Snowflake has announced new technological advancements that will simplify the machine learning (ML) model and full-stack application development processes in the Data Cloud. The latest improvements set to enhance Python capabilities through Snowpark are expected to boost productivity, increase collaboration, and speed up end-to-end AI and ML workflows.
Support for containerised workloads and expanded DevOps capacities will now enable developers to expedite development and run applications through Snowflake's managed infrastructure. The data cloud company's enhancements promise to streamline data exploration and machine learning development for both SQL and Python users. This is achieved through the cell-based programming environment provided by Snowflake Notebooks.
Prasanna Krishnan, the Senior Director of Product Management at Snowflake, stated, "The rise of generative AI has made organisations most valuable asset, their data, even more indispensable. We are making 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 natively in the Data Cloud."
Krishnan went on to highlight the role of Snowflake Marketplace as the first across-cloud marketplace for data and applications in the industry. The marketplace enables quickly and securely implementing what developers have constructed for global end users, thus unlocking increased monetisation, discoverability, and usage.
By introducing extended functionalities, Snowflake is facilitating more complex ML model development and deployment through Snowpark for developers. As of September 2023, more than 35% of Snowflake's customer base was using Snowpark on a weekly basis. Several Snowflake customers, including Cybersyn, LiveRamp, and SNP, have boosted developer productivity by harnessing Snowflake's Native App Framework and unlocked new revenue streams via the Snowflake Marketplace.
Snowflake is also set to debut a new interface - the Snowflake Notebooks - offering an interactive, cell-based programming environment for Python and SQL users to explore data in Snowpark. Furthermore, the new Snowflake Feature Store and Model Registry will allow developers to create, store, manage, and serve ML features for model training and inference, thereby providing a more efficient platform for ML model development.
Saad Zaheer, the Vice President of Data Science and Engineering at Endeavor, a global sports and entertainment firm using Snowflake's Snowpark for Python capabilities to build and deploy ML models, said that Snowpark drives their end-to-end machine learning development. He further noted that it enables them to centralise and process data from various sources and build and train models using that data to create hyper-personalised fan experiences at scale.
Snowflake has also added building blocks to the app development process - the distribution, operation, and monetisation within its platform through the Snowflake Native App Framework, which is expected to increase monetisation opportunities significantly. Another noteworthy feature, the 'Snowpark Container Services', allows developers to run any element of their app without having to move data or manage complicated, container-based infrastructure.
To conclude, the latest developer capabilities unveiled by Snowflake offer comprehensive automated DevOps for apps, data pipelines, and other developments. These features will allow developers to automate key DevOps and observability capabilities across the testing, deployment, monitoring, and operation of their apps and data pipelines. This means they can take their ideas to production faster, ultimately driving increased business impact.