your123 - stock.adobe.com
Domino Data Lab releases latest MLOps platform update
Domino 5.2 includes new features for data scientists to find the best model development environment and use Snowflake by automating model deployment to Snowflake's Data Cloud.
Domino Data Lab, a vendor that provides data science and AI tools for collaboration, on Thursday introduced its latest MLOps platform update: Domino 5.2.
Unveiled on the first day of the vendor's Rev 3 MLOps conference, Domino 5.2 is aimed at data science teams focusing on data preparation and model development.
The release includes new AI-powered data preparation and visualization options. Domino 5.2 includes a new native SQL-based environment based on Apache Superset, an open source platform for visualizing data.
The platform update also provides new AI capabilities that recommend the best model development environment for data science teams, the vendor said.
Another new feature enables data science teams to deploy AI models using Snowflake Snowpark, a platform that enables enterprises to build applications that process data in Snowflake, a cloud data platform vendor.
Customers can train their models using Snowpark and deploy them from Domino to the Snowflake Data Cloud.
A lot of competitors
Domino 5.2 combines the vendor's data science platform and its model monitor product, said Peter Krensky, a Gartner analyst.
The Domino MLOps platform enables data scientists to develop and manage models with their choice of tools and languages, while the model monitor enables organizations to monitor the performance of all their models.
This release helps Domino Data Lab stay competitive in an increasingly crowded field, Krensky said.
Peter KrenskyAnalyst, Gartner
"All of the players that are now becoming either cloud MLOps or cloud end-to-end data science platforms have to compete with each other, which is not easy," he said.
Domino is not only up against other vendors in the MLOps sector, but they're also competing with bigger much bigger vendors including AWS, Microsoft and Google.
So an all-in-one MLOps system ought to appeal to enterprises, Krensky said.
Organizations might be particularly drawn to Domino 5.2's collaboration capabilities that support multiple personas such as data scientists, machine learning engineers, AI architects and engineers, Krensky continued.
Enterprises might also be attracted to Domino 5.2's management of a series of MLOps capabilities, including governance, monitoring and bias detection.
In addition, a platform that helps users raise the percentage of models that make it into production and manage those models is appealing to enterprises, Krensky said.
Helping to solve challenges
However, organizations face significant challenges working with data and MLOps, he continued.
"Data quality is always a major issue," Krensky said. "And the most important parts in data science and MLOps are the least sexy," including the ability to break down data silos. Another challenge in the market is retaining skilled data professionals, Krensky noted.
Until recently, recruiting data scientists was the difficult part of assembling an effective data team. Now the big challenge is getting those recruits to stay, he said.
"This retention problem comes [with] things like traceability, auditability," Krensky said. "Can someone pick up where somebody else left off, but also if there's a regulatory concern or an internal policy review, can you show your work?"
This can affect team continuity, which the Domino MLOps platform is designed to help with, Krensky said.
The platform does this by enabling data science teams to automatically set up prediction data capture pipelines, monitor models deployed to Snowflake Data Cloud and continuously update the data drift and model quality calculations.
Domino did not release pricing information for the subscription-based system. The platform will become generally available to customers in June.