JFrog today revealed it has acquired Qwak to add a machine learning operations (MLOps) platform to its existing portfolio of DevOps tools and platforms.
.
Data science teams typically rely on an MLOps platform to manage the workflows used to build and deploy an artificial intelligence (AI) model and JFrog has already integrated its core DevOps platform with the MLOPs platform created by Qwak.
JFrog CTO Yoav Landman said it’s now only a matter of time before MLOps and DevOps processes converge, as more applications are infused with AI models. The acquisition of Qwak gives JFrog a mature MLOps platform that, for example, provides versioning and immutability capabilities needed to advance that convergence, he added.
Despite the acquisition of Qwak, JFrog will continue to provide integrations with other MLOps platforms such as AWS Sagemaker and MLflow from DataBricks. However, as application development continues to evolve, more organizations are going to require seamless integration of MLOps and DevOps across their software supply chain, noted Landman.
Many of the MLOps workflows that data science teams created replicate many of the same processes already used by DevOps teams. For example, a feature store provides a mechanism for sharing models and code in much the same way DevOps teams use a Git repository. Over time, organizations will need to either integrate these repositories or replace them as workflows become more integrated.
Of course, there will also be significant cultural challenges that will be encountered as organizations look to meld MLOps and DevOps teams. Many DevOps teams deploy code multiple times a day. In comparison, data science teams require months to build, test and deploy an AI model. Additionally, once deployed an AI model tends to drift as additional data is analyzed. That drift can result in algorithms making recommendations and decisions that over time can become increasingly suboptimal. Data science teams, as a result, need to update AI models using workflows that need to be incorporated into a DevOps workflow.
Savvy IT leaders should take care to make sure the current cultural divide between data science and DevOps teams doesn’t get any wider. After all, it’s not so much a question at this juncture whether DevOps and MLOps workflows will converge as much as it is to when and to what degree. The longer that divide exists, the greater the inertia that will need to be overcome to bridge it becomes.
There are, of course, plenty of providers of MLOps platforms that have a vested interest in justifying investments in their platforms, but, at a time when organizations are under more economic pressure than ever to reduce costs, there may be no better time than the present to identify a set of workflows that are increasingly become redundant to one another. The fact of the matter is building, updating, securing and deploying AI modes is a repeatable process that lends itself to automation, and there are already more than a few data science teams that would prefer it if someone else in the IT organization managed this process on their behalf.
