How does the support process occur: through git or manually?

The process of supporting machine learning models is usually organized through version control systems such as Git, rather than manually. This allows:

• Tracking changes in code, configurations, and data.
• Restoring previous versions of models and experiments.
• Collaborating within a team.

In addition to code, it is important to version data and models for ML projects. Specialized tools (DVC, MLflow, etc.) are used for this purpose, which integrate with Git or operate in parallel.

An example workflow:

1. Developing and training the model with code and configurations committed to Git.
2. Versioning data and models using DVC.
3. Automating deployment and monitoring through CI/CD.

Thus, supporting ML projects is a comprehensive process based on version control systems and specialized tools, rather than manual management.