An MLOps team implements automated model retraining triggered by data drift detection. From a model risk management perspective, what control is MOST critical to include in this retraining pipeline?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — you need an independent validation gate before a retrained model enters production because automated retraining still counts as a material model change under model risk management. Skipping that review lets drift-driven changes slip into production without proper oversight.
Full Explanation
An independent validation gate before the retrained model enters production is correct because SR 11-7 and model risk management best practices require material model changes, including retrained models, to undergo independent validation before production deployment, even when retraining is automated. Automated deployment without human review to minimize latency bypasses controls and introduces unacceptable risk. End-user notification to all users that the model has been updated is a communication practice, not a primary control. Retrained model compression is a performance optimization, not a risk control.