Gets the data into a shape a model can actually use, reliably and repeatedly.
Why this one is hard to judge
Most AI projects fail here rather than at the model, and the people who know that are rarely the loudest in the room.
What to ask for
Ask for a pipeline they built and what happens when the source data changes shape.
Ask how they would find out that a dataset silently went stale.
Ask for a specific data quality issue that reached a model before anyone noticed.
Ask what they would do differently if storage cost had to fall.
The mistake most hirers make
Hirers focus on the volume of data someone has handled rather than how they noticed when it was wrong. A pipeline that runs quietly and produces bad data is more damaging than one that fails loudly. Ask what tells them the data has gone bad, not just how much of it they moved.
What good looks like after 90 days
A pipeline with a check that would have caught the team's last data problem. A written map of where important data comes from and what depends on it. One source of silent data drift identified and monitored.
How we assess it
Against a rubric that is published in full, on evidence the practitioner supplies and a reviewer checks. Where something has not been verified, the profile says so.