Yes, but kata solutions don't necessarily have to be efficient (unless the kata tests explicitly require them to be such).

merged, thanks!

The goal is to find the specialists who are doing mistakes in employee names.

So you have to return the specialist id, not the employee id. As written in the kata description:

Resultant columns:

1. specialist_id (int): The unique ID of the specialist (certifications.entered_by).

I hope that helps.

Fix description

1. Removed description of employees.department field which doesn't exist.

2. Reordered fields description of certifications table to match the database.

3. Fix typo in the resultant table field name error_percentage.

Fix: forbid the import of pathlib.

Fix: swap Expected and Actual. Fixes this issue.

Fix: swap Expected and Actual.

This is so elegant.
It would get better rating in the "Best Practices" if you put the variable in quotes.

Super clean and readable!