Skip to content
Skills
Problems
Questions
Roles
Skills
Interview
Contest
Leaderboard
More
Problems
Questions
Roles
Skills
Interview
Contest
Leaderboard
More
Home
/
MLOps Engineer
/
Questions
/
ML pipelines and reproducibility
MLOps Engineer
ML pipelines and reproducibility questions
20 questions on ML pipelines and reproducibility for mlops engineers, easiest first.
Easy
7
What a model registry is for
easy
Why version data
easy
What experiment tracking records
easy
Same code, different results
easy
Logging a metric every epoch
easy
Making a training run repeatable
easy
Bad data reached training
easy
Medium
6
When to retrain
medium
Online and offline feature stores
medium
Safe retries in pipelines
medium
Joining features to training labels
medium
Deploying by alias, not version
medium
Online features stopped updating
medium
Hard
7
Rollback that made things worse
hard
Same seed, different results
hard
Loading a model from a public hub
hard
Daily batch missing late events
hard
Pipeline reused yesterday output
hard
Streaming counts that run high
hard
Loss spike after resuming training
hard
