DevML and Not MLOps

This blog is my point of view about why DevML or MLDev is a better term to represent Machine learning developers than MLOps due to the fact that the machine learning developers don't understand the operations part of Ops in DevOps.

The term "DevOps" has been gaining a lot of traction lately, and for good reason. It's a powerful concept that can help organizations improve their agility and speed of delivery. However, when it comes to machine learning, the term "DevOps" doesn't quite fit.?

First of all, machine learning developers don't have a strong understanding of operations because they're focused on building models that work well and are scalable. DevOps is all about ensuring that machine learning models are deployed and running smoothly, so it's understandable that developers would be less focused on that area. However, it's important for developers to have at least a basic understanding of operations so that they can build models that are both effective and scalable.

Another reason why the term "MLOps" doesn't quite fit when it comes to machine learning is that machine learning is a process that is constantly evolving. As new data is collected and new algorithms are created, the machine learning system needs to be updated on a regular basis. This process is different from traditional software development, which follows a more linear model.

In traditional software development, you have a linear process where you first gather requirements, then design the system, implement it, and test it. With machine learning, you still have to gather requirements and design the system, but the implementation and testing are done through building models. This iterative process allows for more flexibility and creativity in the development process.

So, while "DevOps" is a great term for describing the collaboration between developers and ops teams, it doesn't really capture what machine learning developers do.