A central challenge that organizations face when enabling new machine learning use cases, is that they struggle to scale the amount of self-developed ML models in production due to a high rate of project failure. This is due to not only technical, but also organizational challenges, such as:
- Lack of detailed guidelines and responsibilities to define the right team composition
- Machine learning projects are inherently interdisciplinary
- The evolving skill requirements throughout the lifecycle of a machine learning project
In this whitepaper, we introduce the Machine Learning Skill Profiles framework as a comprehensive organizational blueprint to scale machine learning in enterprises.
We identified ten different roles that contribute to a machine learning project throughout its lifecycle. We name these skill profiles and discuss their organizational embedding, responsibilities, skills, and educational requirements. This results in benefits such as separation of responsibilities, effective project staffing and long-term upskilling and hiring plans.
Finally, we provide thoughts on how to apply the framework to your company, based on company size, the level of AI maturity, and the desired degree of centralization. With these insights, we propose a framework that helps managers and practitioners who are building ML teams in their organizations do so more effectively.
The report is the result of the appliedAI MLOps working groups. It is based on the experience of leading experts from appliedIA partner companies.
Authors of the Whitepaper:
- Alexander Machado, Head of Trustworthy AI at appliedAI Initiative
- Max Mynter, MLOps Engineer at appliedAI Institute for Europe
We thank you for your contributions
- Elena Zennaro (Infineon)
- Dr. Hendrik Brakemeier (European Central Bank)
- Salma Charfi and Eduard Götmann (Miele)
- Benjamin Pohl (EnBW)
- Mark Mauerwerk and Natalia Fitis (Deutsche Telekom)
- Matthias Berger (MTU Aero Engines)
- Jonas Goltz (Giesecke+Devrient GmbH)
- Tobias Emrich (Snke OS)
and many more.