Which activity belongs to the deployment stage of the machine learning lifecycle?
Show answer and explanation
Answer: B
Deployment takes a validated model and makes it available to real consumers, typically behind an endpoint, a batch job or an embedded component. It includes packaging dependencies, choosing compute, configuring authentication, and setting up logging and monitoring. Deployment is not the end of the lifecycle: performance is watched for drift, and retraining loops feed improvements back into development. Data collection and algorithm selection belong to earlier stages.
Why the other options are wrong
- A. Collecting examples is part of data acquisition at the start of the lifecycle.
- C. Cleaning and shaping data is data preparation, which precedes training.
- D. Algorithm selection happens during experimentation, before a model is deployed.
Exam tip. Deployment equals serving predictions.