A late-arriving event stream is poisoning 30-day aggregations. How do you fix the pipeline?
PICTURE THIS: DATA SPLIT
Fit on train, tune on val, report on test once.
Simple meaning
Use event-time watermarks, allowed lateness, and versioned re-aggregation windows rather than processing-time counts.
WHY — Feature Pipelines instead of guessing?
Why interviewers care about Feature Pipelines:
people who only read docs from people who shipped.
and tied to MLOps work.
Name the idea, why it exists, then one short example.
End with when you use it and one common pitfall.
STEPS — What happens step by step?
Before you speak the answer, walk the interviewer through these steps:
- 1Use event-time watermarks, allowed
lateness, and versioned re-aggregation windows rather than processing-time counts.
- 2Backfill the offline store
and only then rematerialize online keys.
- 3Document that training will
ignore incomplete windows until the watermark closes.
- 4Give an example
One tiny concrete case you can say aloud.
- 5Common mistake
What juniors usually get wrong.
- 6Close
When you pick this over the alternative.
EXAMPLE — See it in action
Here's a short line you can speak, broken into clear beats:
Note: Adapt this scaffold to your own project — keep it under 60–90 seconds.
Key takeaway
Use event-time watermarks, allowed lateness, and versioned re-aggregation windows rather than processing-time counts. Backfill the offline store and only then rematerialize online keys.