Customized MLOps solutions for the industry

Mastering the challenges of tool selection and integration

© Have a nice day /

There are many commercial and open source tools available in the market for developing ML solutions within the industry. However, the process of selecting a specific tool is not straightforward. This complexity arises from various use-case-specific constraints, such as licensing agreements, data security, and compatibility with existing systems etc . Moreover, multiple tools need to be combined to achieve the desired functionality. This introduces challenges in terms of usability, operability, maintenance, and integration into the existing infrastructure of industry partners.

Our researchers in the Center for Applied Research on Supply Chain Services at Fraunhofer IIS are dedicated to developing custom machine learning operations (MLOps) solutions that are tailored towards the specific requirements of our industry partners. These MLOps solutions leverage the strengths of various ML specific tools while still being easily operable and maintainable.

Procedure for the development of customized MLOps solutions

First, we analyze requirements through in-depth interviews with relevant stakeholders in the company. Through extensive research, we identify and evaluate suitable tools in accordance with the use case requirements. The goal is to combine all identified tools and develop a custom MLOps solution tailored to the current use case. This solution includes all the necessary functionalities essential for an ML solution, as illustrated in the architecture diagram in Figure 1.

© Fraunhofer IIS
ML architecture diagram describing various components needed to develop an ML application.

Subsequently, this MLOps solution can be integrated into the customer's infrastructure and is complemented by a catalog summarizing the conducted interviews and the recommendations derived from our research.

If you are looking for custom MLOps solutions for your company, we are here to assist you. Please feel free to contact us.

You might also be interested in


Machine Learning Operations (MLOps)

Find out more about how you can manage AI systems efficiently.


Cutting-edge data analytics methods

Extracting the unique assets in companies’ data using analytics methods and artificial intelligence


Data Efficient Automated Learning

Read more about our competencies for more efficiency of AI within the Machine Learning Lifecycle here.