Machine Learning becomes more available

[BeauV]

My last company was deeply involved in the development and expansion of something called Kubernetes, which is a way of managing large numbers of things called containers across farms of servers and keeping the code in those containers running despite the occasional server or even data center failure. As a result, I'm still fascinated by the work-loads which are appearing on that software platform.

This latest is a really interesting approach to prying the true Machine Learning bit away from all the supporting infrastructure stuff. You can read about it HERE.

The amazing thing is that software platforms like this were highly secretive in the past and served as the financial under pinnings of a number of large companies. Yet, as with so many other software bits, as the industry standardizes on a workload and approach, the open source software community hollow out the proprietary platforms by offering free versions that are good enough. I hadn't expected Machine Learning to follow this path quite this quickly.

[BeauV]

Here's another example: OpenAI

[Soñadora]

In the next few months, we will be deploying Kubernetes to manage our coalesce our Docker containers. This is new territory for us and a very cool abstraction of the typical discrete service deployment in SOA.