What it is MLOps ? ML + Ops ?
The term MLOps start appearing a lot recently in articles and tweets, but what does it actually means?
I will be following the way we used in college classes to explain complicated terms combined of multiple known words. we split and define each of the words and explain it!
MLOps is combined of two words :
📌 ML : Machine Learning is a set of algorithms and techniques that can systems the ability to learn and improve from data. By extracting patterns that can’t be extracted easily “people works on ML are called : Data scientists ML engineers”
📌 Ops : IT Operations are the sets of processes , methodologies and technologies that aims to shorten a system lifecycle. By reducing and guaranteeing the delivery of the system into production “people works on ML are called : Operations Professionals”
Now that we have defined the 2 main parts of the MLOps lets try combine everything together and talk about MLOps meaning, why do we even need it and what are pre-requirements to build MLOps!
📌 MLOps : Machine Learning Operations is the collaboration between ML specialists and Operations professionals in order to define best practices/discipline to increase the quality, reliability and efficiency of ML systems from Data collecting to Deployments and Monitoring.
Moving away from the academic way of defining the term. MLOps can be also seen as rule of 25% by saying that MLOps is 4 main parts that are equally important when we are talking about MLOps.
“software engineering, data engineering, modeling, and the business problem”
Now that we have an idea about MLOps meaning, we would ask why do we even need such a discipline?
📈 we can find our answer in NewVantage-Partners-Survey “91.5% of companies reporting ongoing investment in AI. Only 14.6% have deployed AI into production”
https://bwnews.pr/3rgmjpm
If we can describe the idea of human needs by Maslow’s hierarchy the same concept goes through MLOps, in order to achieve MLOps we need first to have DevOps, Data automation and platform automation best practices.
📈 The next picture is from the book : Practical MLOps ‘Oreilly’
The end of post! Thanks for reaching this far
