Features for training and serving on the Build Your Own AI Platform track. A feature used at training must be computable the same way at serving. A feature store, even a plain table with definitions, is that contract.
This lesson assumes you already worked through Batch inference.
The idea in practice
Document point-in-time correctness: the feature value as of the decision, not a future aggregate.
A concrete check
goal = {
'track': 'Build Your Own AI Platform',
'lesson': 'Features for training and serving',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
A training column that is recalculated differently in the API. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
Pick one feature and write the training definition and the serving definition. They must match.
Self-check
- Say Features for training and serving in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.