Model Life-Cycle Management at Continental Tires

This paper discusses the journey and lessons learned in managing the life cycle of machine learning (ML) and artificial intelligence (AI) models as Continental Tires industrialized its data science initiatives.

Key points covered include:

1. The importance of a bottom-up approach and establishing a common infrastructure for all data science teams.

2. The need for collaboration between data scientists and IT teams to develop processes and bridge gaps between business and IT departments.

3. The process of industrializing models and creating data products that optimize business processes or extend existing products.

4. The essential elements of model life-cycle management, including data and feature versioning, model versioning, model warm-up, model deployment, and model retraining.

5. The significance of the model warm-up phase in testing the model's performance in real-world scenarios and the need for a protected environment that allows data scientists to "fail fast, learn fast."

6. The deployment of models and the distribution of responsibilities among different teams to maintain and improve the software.

7. The importance of enabling the data science community by providing a provisioning environment and promoting data-driven work across the company.

The paper concludes by highlighting the success of Continental Tires in developing AI-based products for internal process optimization and the strong foundation they have established for quickly evolving new use cases and rolling them out as productive software.

What's inside

• Collaboration — Bridging gaps between business and IT for successful AI model development and deployment.
• Industrialization — Creating data products that optimize processes and extend existing offerings.
• Life-Cycle Management — Comprehensive approach to manage models from versioning to deployment and retraining.
• Enablement — Empowering data scientists with provisioning environments and promoting data-driven work.