Over the last few years, we’ve seen a dramatic increase in enterprise adoption of artificial intelligence (AI) and machine learning (ML). Enterprise organizations in every industry are investing in AI / ML for a wide range of uses cases to drive innovation, achieve competitive advantage, and deliver business value.
This early adoption focused on experimentation with a few targeted proof-of-concept projects. Now enterprises are moving beyond the experimentation phase; they are looking to operationalize their ML models across their business processes. However, a vast majority of them stumble at the “last mile” of model deployment and management, and fail to realize business value with their ML models. These “last mile” issues are typically caused by a lack of tools and standardized processes necessary to facilitate the operationalization of machine learning models.
We have seen this situation play out before. In the pre-DevOps days, software engineering teams were running into similar issues with application development projects and experienced significant delays in moving from development and testing to production. DevOps introduced standardized processes that have made software releases more reliable, bringing speed and agility to the software development lifecycle.
There are some similarities between the work of software developers and that of data science teams, so some aspects of DevOps can be used to address the operationalization of ML models. However, there are also several differences. The machine learning lifecycle (from data preparation, to building and training models, to inference and deployment, to monitoring) is a very iterative and intricate process. Force-fitting DevOps processes onto the machine learning lifecycle isn’t the right approach.
To address this growing need in the market, Hewlett Packard Enterprise (with solutions from BlueData) is introducing the new HPE Machine Learning Ops (HPE ML Ops) solution. HPE ML Ops is a software solution that extends the capabilities of the BlueData container platform and brings DevOps-like speed and agility to the machine learning lifecycle.
We are proud of the ongoing partnership with our long-standing partner and early BlueData investor, Intel. Intel and the BlueData team at HPE share a joint commitment to helping our enterprise customers succeed in their AI journey.