The engineering of AI systems requires the management of many integral artifacts. These include inputs such as data sets, configurations such as hyperparameters and outputs such as training results. However, the methodological approach for such management is often unclear in practice.

In this training course, fortiss researcher Alexandros Tsakpinis will give you an introduction to AI engineering - moving away from the prototypical environment within a Jupyter notebook - and show you how the aforementioned components of AI systems can be systematically versioned and configured.

In a combination of short theory units and live coding sessions, you can get to know open source technologies for a standardized project structure, data versioning and experiment tracking. There will also be the opportunity to discuss emerging questions and real problems from your company. The training will be held as part of the Mittelstand-Digital Zentrum Augsburg.

Contents

• Comparison between the development of an AI prototype in Jupyter Notebook and a standardized project structure
• Methods for data versioning and experiment tracking
• Overview of open source technologies for a standardized project structure, data versioning and experiment tracking

Benefits for the participants

• You recognize the advantages, such as traceability, that arise from versioning data, models, and code.
• You can transfer and apply the examples from the live coding sessions in your company.

This is a follow-up training to the webinar "Best Practices for Developing and Managing AI Projects" on January 24. Participation in the webinar is not required for the training. Read more

Registration

Register now for the free training!

Register now!