Towards practicable Machine Learning development using AI Engineering Blueprints

Nicolas Weeger, Annika Stiehl, Jóakim von Kistowski, Stefan Geißelsöder, Christian Uhl · 2025

Read Towards practicable Machine Learning development using AI Engineering Blueprints by Nicolas Weeger, Annika Stiehl, Jóakim von Kistowski, Stefan Geißelsöder, Christian Uhl (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-06391.txt.

Introduction

Nicolas Weeger and co-authors set out a research plan, published in 2025, to make machine learning projects easier for small firms. Big firms have teams and pipelines for building and running AI, a craft called AI engineering. Small and medium firms often lack that staff, so promising models stall before use.

The plan develops blueprints: reference designs plus automation recipes for common kinds of machine learning systems. Each blueprint is tried in real field projects, and lessons feed back into revised versions. The team judges success by watching development work and interviewing the builders. If the blueprints hold up, a small firm could assemble a sound AI system from proven parts instead of hiring scarce experts. The paper frames the studies still to come.

Catalog record

CREATORNicolas Weeger, Annika Stiehl, Jóakim von Kistowski, Stefan Geißelsöder, Christian Uhl
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE16 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-06391

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Towards practicable Machine Learning development using AI Engineering Blueprints Nicolas Weeger, Annika Stiehl, Jóakim von Kistowski, Stefan Geißelsöder, Christian Uhl Abstract The implementation of artificial intelligence (AI) in business applications holds considerable promise for significant improvements. The development of AI systems is becoming increasingly complex, thereby underscoring the growing importance of AI engineering and MLOps techniques. Small and medium-sized enterprises (SMEs) face considerable challenges when implementing AI in their products or processes. These enterprises often lack the necessary resources and expertise to develop, deploy, and operate AI systems that are tailored to address their specific problems. Given the lack of studies on the application of AI engineering practices, particularly in the context of SMEs, this paper proposes a research plan designed to develop blueprints for the creation of proprietary machine learning (ML) models using AI engineering and MLOps practices. These blueprints enable SMEs to develop, deploy, and operate AI systems by providing reference architectures and suitable automation approaches for different types of ML. The efficacy of the blueprints is assessed through their application to a series of field projects. This process gives rise to further requirements and additional development loops for the purpose of generalization. The benefits of using the blueprints for organizations are demonstrated by observing the process of developing ML models and by conducting interviews with the developers. I Introduction Artificial intelligence (AI) is transforming numerous industries and application domains. It is crucial for organizations to adopt AI techniques in order to achieve business success [ 1 , 2 ] . SMEs in particular can benefit significantly from adopting AI techniques. They can improve or even establish capabilities in certain areas, such as customer experience, production monitoring, and decision-making processes [ 3 ] . The proprietary development of ML models for use as or within a product, called AI systems, in an organizational context can lead to a number of challenges [ 4 , 5 , 6 , 7 ] . This includes understanding of the intricacies of AI, including its functional requirements and operational scenarios. The integration of additional processes, such as data generation and preprocessing or model training and deployment, with the traditional software engineering process can potentially create organizational constraints, particularly for SMEs [ 6 ] . The ML model development lifecycle includes additional practices beyond DevOps for the data and models. These include MLOps and DataOps techniques that provide a culture, practices, and tools for handling data and models. In addition, the system architecture for AI systems must be aligned with the requirements of the underlying model. Training and inference environments, along with data storage and versioning of the different artifacts, must be incorporated in order to enable the system to function as an AI system. Consequently, the effective implementation of AI systems requires a well-designed architecture that is tailored to the specific requirements of the intended AI application. This paper discusses the importance of AI engineering practices and the concept of developing blueprints tailored to the requirements of different types of AI and stages of development. We will discuss the content of these blueprints and how they can be utilized by organizations. The discussion will commence with a preliminary result in the form of a pipeline and its sub-pipelines. The results of this research will assist organizations to address these challenges and streamline the development, deployment and operation of AI systems. Consequently, they will be enabled to adopt the blueprints by providing reference architectures and suitable automation approaches for various types of AI. II Background and Related Work II-A AI Engineering AI engineering is an evolution of the field of software engineering and, due to the rapid growth of ML developments, it is an emerging field. AI engineering is currently at the maximum ‘‘peak of inflated expectations”, as evidenced by Gartner’s AI Hype Cycle for 2024 1 1 1 https://www.gartner.com/en/articles/hype-cycle-for-artificial-intelligence . According to Gartner, “AI engineering is the foundation for enterprise delivery of AI and GenAI at scale. Most organizations lack the data, analytics and software foundations to move individual AI projects to production at scale - much less operate a portfolio of AI solutions at scale.” Over a dozen projects were examined in [ 8 ] , where AI engineering challenges led to problems with productizing ML models. The study found that the majority of companies that develop machine learning models encounter difficulties when trying to move them into production. They provide a research agenda and overview of the issues that need to be addressed in this direction. [ 9 ] point out that while the field of software engineering research has been extensively discussed, AI engineering has been much less addressed. Only a limited number of publications present concrete experiences related to the application of AI engineering principles. They selected ten AI engineering practices in several categories from the literature and applied them to an example implementation to evaluate the practices and their systems architecture. Furthermore, discussions and questionnaires, especially with SMEs, have shown that there is a desire to implement AI in their systems. However, the success of their model implementation and productization depends on overcoming the challenges mentioned above. Thus, applying AI engineering practices can help organizations streamline the development, deployment and operation of machine learning models. II-B MLOps The idea of MLOps is to provide techniques and tools for the deployment and operation of AI systems [ 10 ] . The goal is to devise a strategy for solving real-world problems with the deployment of ML models. Several studies review various literature in this area and offer pipelines, taxonomies, tools, methods and challenges in this area [ 11 , 12 , 13 ] . [ 14 ] conduct a systematic mapping study for MLOps architectures and point out 35 architecture components, describe different architecture variants for different use cases and provide popular tools to use for these architecture components. II-C Other Related Work In [ 15 ] a reference architecture for the specific use cases in the process industry dealing with edge devices was presented. They demonstrated the architecture by implementing a case study for a real-world use case and proved the functionality with this application. [ 16 ] developed a reference architecture to facilitate the use case of big data in edge computing ML techniques. They come up with different views on the architecture of model development and deployment for this specific use case. Another study [ 17 ] presents a vision for “disciplined, repeatable and transparent model-driven development and Machine-Learning operations (MLOps) of intelligent enterprise applications.” They provide a three-stage metamodel for model based development of AI/ML blueprints based on intelligent application architecture. With scope on software and architecture, design patterns for AI based systems are discussed in several studies [ 18 , 19 , 20 , 21 ] . They provide an overview of design patterns, adapted or newly generated for AI use cases, and show the application and the resulting benefits of using them to develop machine learning models. In summary, the literature provides insights into the importance, potential architectures, and principles for AI engineering and MLOps practices. However, the application of these insights is currently focused on a few reference architectures in specific domains, such as big data or edge devices. Other studies

Read the complete text