Moderate Training-Serving Skew Question 130 of 221

What are common sources of training-serving skew besides code duplication?

MLOps track · Speak this in 60–90 seconds · Faridabad & Delhi NCR

PICTURE THIS: DATA SPLIT

Train 70%Val 15%Test 15%

Fit on train, tune on val, report on test once.

Simple meaning

Clock skew in timestamps, different timezone handling, online features that are not backfilled, and batch aggregations that include the label window.

WHY — Training-Serving Skew instead of guessing?

Why interviewers care about Training-Serving Skew:

Training-Serving Skew questions separate

people who only read docs from people who shipped.

Keep it short, concrete,

and tied to MLOps work.

Stay structured

Name the idea, why it exists, then one short example.

Close cleanly

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:

1
Clock skew in timestamps,
different timezone handling, online features that are not backfilled, and batch aggregations that include the label window.
2
Library version drift between
train image and serve image also counts.
3
Logging both feature vectors
helps you catch it.
4
Give an example
One tiny concrete case you can say aloud.
5
Common mistake
What juniors usually get wrong.
6
Close
When you pick this over the alternative.

EXAMPLE — See it in action

Here's a short line you can speak, broken into clear beats:

Say this line

“Library version drift between train image and serve image also counts.”

Break into beats

Libraryversiondriftbetweentrainimage

Speaking order

2987408337471632900

Note: Adapt this scaffold to your own project — keep it under 60–90 seconds.

Key takeaway

Clock skew in timestamps, different timezone handling, online features that are not backfilled, and batch aggregations that include the label window. Library version drift between train image and serve image also counts.