High CI/CD for ML Question 153 of 221

How do you test a non-deterministic model in CI without flaking the build?

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

Fix seeds where possible, compare metrics with tolerances, and assert ranking or calibration bands rather than exact floats.

WHY — CI/CD for ML instead of guessing?

Why interviewers care about CI/CD for ML:

This is a process

question about CI/CD for ML.

Panels listen for order,

trade-offs, and what you would actually do on a MLOps project - not buzzwords.

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
Fix seeds where possible,
compare metrics with tolerances, and assert ranking or calibration bands rather than exact floats.
2
Use a golden set
of relative checks such as better than a dummy.
3
Flaky GPU jobs belong
on a retrying nightly pipeline, not on every lint push.
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

“Use a golden set of relative checks such as better than a dummy.”

Break into beats

Useagoldensetofrelative

Speaking order

2987408337471632900

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

Key takeaway

Fix seeds where possible, compare metrics with tolerances, and assert ranking or calibration bands rather than exact floats. Use a golden set of relative checks such as better than a dummy.