You recently joined a machine learning team that will soon release a new project. As a lead on the project, you are asked to determine the production readiness of the ML components. The team has…

Question

Options

• AEnsure that training is reproducible.
• BEnsure that all hyperparameters are tuned.
• CEnsure that model performance is monitored.
• DEnsure that feature expectations are captured in the schema.

• A
  14% (9)
• B
  6% (4)
• C
  77% (51)
• D
  3% (2)

Explanation

Monitoring is the missing pillar in this team's readiness checklist. ML production readiness frameworks (like Google's ML Test Score) organize checks into four areas: features/data, model development, infrastructure, and monitoring. Since the team has covered the first three, the gap is production monitoring - specifically, tracking model performance over time to catch issues like data drift, concept drift, or degraded accuracy after deployment.

Why the distractors are wrong:

• A (reproducible training) belongs to the model development category, which the team already tested.
• B (hyperparameter tuning) is also a model development concern - optimization happens before release, not as a readiness gate.
• D (feature schema/expectations) falls under features and data testing, already completed.

Memory tip: Picture the four readiness pillars as a table with legs: Data, Model, Infrastructure, and Monitoring. The question tells you three legs are in place - the table will fall without the fourth (Monitoring). When you see "already tested X, Y, Z," map those to the framework and identify the missing category.

No community discussion yet for this question.