Top 7 Mlops Interview Questions
Browse 7 real-world technical and behavioral interview questions about Mlops. Review scenarios, edge cases, and architectural best practices.
Is a feature store worth its operational cost, or can you argue against one?
It earns its cost when several models share features and when you need offline and online parity plus feature history for correct backfills. It does not when one team runs one model, because you have added a stateful service to the request path for benefits nobody is drawing on yet.
Where does train-serve skew come from, and how do you stop it?
Train-serve skew is a mismatch between features used in training and features produced at serving time. Prevent it with shared transformations, point-in-time joins, feature logging, schema checks and monitoring of served feature distributions. It also connects feature store to the point an interviewer is testing.
A model scored well in validation and performs badly in production, with no errors anywhere. What do you check first?
Check whether training and serving compute the same features from the same definitions, because the usual cause is two implementations that disagree slightly. Nothing raises an error when they diverge - the model receives plausible numbers that mean something different from the ones it learned on.
What is a model registry for, beyond storing the files?
It is the system of record for which artefact is authoritative and why. It binds each version to the data snapshot, code commit and evaluation that produced it, records stage transitions and who approved them, and gives serving a stable identifier so any past prediction can be traced to the exact model that made it.
What has to be captured for a training result to be reproducible?
Five things: the data version, the code commit, the resolved configuration, the pinned environment and the seeds. Even with all five, GPU kernels are often nondeterministic, so what you can promise is a statistically equivalent rerun and a recoverable artefact, not a bitwise identical one.
What has to be in place before you would let a model serve live traffic?
Features must be computable at decision time, training and serving must share one transformation implementation, the model and dataset version must be pinned together, drift and realised performance monitored separately because labels arrive late, and rollback must move model and features as one unit.
Why isn't a notebook a record of an experiment?
A notebook records the cells you wrote, not the order you ran them or the state they ran against, so the saved file frequently cannot reproduce its own outputs. Keep notebooks for exploration but promote anything you intend to defend into a parameterised script that logs its inputs to a tracking store. It also connects experiment tracking to the point an interviewer is testing.