Practical MLOps

Chapter 8. MLOps for Azure

By Alfredo Deza

A third reason for our family to move was that we had never had a proper home since 1933 when, in the midst of the Great Depression, we were dispossessed. It was only years later that I understood how the rural paradise I loved at age 6 disappeared. My parents could not make the payments. My mother abandoned her struggling practice and found a job as receiving physician in a state hospital that provided a few dollars and an apartment too cramped for us all, so my brother and I were sent into what we later called “The Exile.”

Dr. Joseph Bogen

Microsoft’s continuous investments in Azure for machine learning are paying off. The number of features offered today makes the platform as a whole a great offering. A few years ago, it wasn’t even clear that Azure would get such an influx of top-level engineering and a growing interest in its services.

If you haven’t tried Azure at all or haven’t seen anything related to machine learning within Microsoft’s cloud offering, I highly recommend you give it a chance. Like most cloud providers, a trial period is available with enough credits to try it out and judge for yourself.

One of the examples I usually tend to bring up is that of using Kubernetes. Installing, configuring, and deploying a Kubernetes cluster is not a straightforward task at all. It is even more complicated if you factor in associating a machine learning model and scaling its interactions with potential consumers. This is a challenging ...