PMLE Collaborating to manage data and models Practice Question
Your organization uses Vertex AI Feature Store to serve features for a real-time fraud detection model. Multiple teams contribute features, and you need to ensure that feature values are consistent between training and serving. Which practice should you implement to prevent training-serving skew?
⚠ Common exam trap
The trap here is thinking that monitoring and retraining can solve training-serving skew, but they only detect and react to it; the only way to prevent it is to ensure identical feature computation for training and serving.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Use a single featurestore for both training and serving, and ingest features via the same pipeline.
Training-serving skew arises when feature values differ between training and serving due to disparate data processing. Using a single featurestore and a unified ingestion pipeline ensures that features are computed and stored consistently, so the model sees identical feature distributions in both phases. This is a fundamental best practice in ML engineering.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Implement a separate feature engineering pipeline for training and another for serving to optimize each for its specific needs.
Why it's wrong here
Separate pipelines often lead to divergent logic, such as different handling of missing values or categorical encoding, which causes training-serving skew. While optimization is possible, the risk of inconsistency outweighs benefits. A single pipeline ensures identical transformations, which is critical for model performance.
- ✗
Export features from the featurestore to BigQuery for training, and use the featurestore online serving for predictions.
Why it's wrong here
Exporting features to BigQuery for training while serving from the featurestore can introduce skew if the export process applies different transformations or if data freshness differs. This decouples training and serving pipelines, increasing the risk of inconsistency. A unified approach is required to prevent skew.
- ✓
Use a single featurestore for both training and serving, and ingest features via the same pipeline.
Why this is correct
Using a single featurestore and a unified ingestion pipeline ensures that the same transformation logic and data sources are used for both training and serving, eliminating discrepancies. This practice directly addresses training-serving skew by maintaining consistency in feature computation and storage, which is essential for model reliability.
- ✗
Use Vertex AI Feature Store's built-in monitoring to detect skew, and retrain the model when skew is detected.
Why it's wrong here
Monitoring can detect skew after it occurs, but it does not prevent it. Retraining may mitigate symptoms but does not address the root cause of inconsistent feature computation. Proactive prevention through unified pipelines is more effective than reactive monitoring and retraining.
Go deeper
Related to this question
About these practice questions
This PMLE question is part of Courseiva's 775-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Google Cloud exam blueprint
This PMLE practice question is part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the PMLE exam.