To remove dependence on mlflow, we should create a new database table to capture metrics associated with a job.

A job can have multiple metrics with the following properties:

• name (str)
• value (float)
• step (int)

The service layer should be updated to store this information int he new table instead of using the mlflow client.