MLOps solution design for a manufacturing company

Many companies today are trying to efficiently use data that is often generated in huge volumes inside and outside their enterprise. One of the first steps is often to store data and analyze it using business intelligence tools. Recently, however, advanced data analytics, so-called data science or the field of machine learning, have become increasingly important. Companies are thus plunging into this field, but often the resulting solutions are rather temporary, unscalable and unsustainable in the long run. One such case will be solved in this paper when these (and other) problems are faced by a real manufacturing company and it needs help with its solution. The aim of this work is to design a solution that meets the requirements and problems of the client, which is a real company operating in the manufacturing sector. The results and benefits of the work are: • acquaint the reader with the field of MLOps and related technologies and principles (containerization, the architecture of microservices, Kubernetes, Kubeflow,), • technological architecture of the solution respecting the requirements of the client, • implementation of the proposed solution in the form of a description of the procedure, including graphical representation (screenshots), • the source code used to implement the solution.