Chuyển đến nội dung chính

Lesson 10: Cheat Sheet & GCP MLE Exam Strategy

Full course summary for GCP Professional Machine Learning Engineer. GCP service reference, evaluation metrics, domain weights, and exam strategy.

📝 Exam Prep — Lesson 10 Lesson 10: Cheat Sheet & GCP MLE Exam Strategy

Google Cloud Professional Machine Learning Engineer Exam Prep

Part 5: Responsible AI & Review

xdev.asia

1. GCP Professional ML Engineer Exam Structure

ItemDetails
Total Questions60 questions
Time Limit120 minutes (2 hours)
Passing Score~70% (Google does not publish exact score)
FormatMultiple choice, multiple select
Validity2 years
LevelProfessional (intermediate to advanced)

2. Domain Weights

DomainWeight
1. Architecting low-code ML solutions~10%
2. Collaborate within and across teams to manage data and models~20%
3. Scale prototypes into ML models~20%
4. Serve and scale models~20%
5. Automate & orchestrate ML pipelines~20%
6. Monitor ML solutions~10%

3. GCP ML Services Cheat Sheet

TaskGCP Service
No-code image classificationVertex AI AutoML Image
SQL-based ML in data warehouseBigQuery ML
Custom TensorFlow/PyTorch trainingVertex AI Custom Training
Hyperparameter optimizationVertex AI Hyperparameter Tuning (Bayesian)
Feature consistency training/servingVertex AI Feature Store
ML workflow orchestration (pipelines)Vertex AI Pipelines (KFP)
Experiment trackingVertex AI Experiments
Model versioningVertex AI Model Registry
A/B testing model versionsVertex AI Endpoints traffic splitting
Monitor feature skew/driftVertex AI Model Monitoring
Explain model predictionsVertex AI Explainability (SHAP, IG)
Real-time event ingestionPub/Sub
Batch + streaming ETL (unified)Dataflow (Apache Beam)
Spark/Hadoop workloadsDataproc
ML pipeline orchestration (multi-service)Cloud Composer (Airflow)
Natural language analysis (no training)Cloud Natural Language API
Document extractionDocument AI
Speech to textCloud Speech-to-Text API
Prevent data exfiltrationVPC Service Controls
Customer-managed encryptionCloud KMS (CMEK)

4. Common Exam Traps

TrapCorrect Answer
"No ML expertise, image classification"AutoML Image (not custom training)
"Train on data already in BigQuery"BigQuery ML (not Vertex AI)
"Features differ at training vs serving"Vertex AI Feature Store (not re-training)
"Trigger retraining when data arrives"GCS notification → Eventarc → Vertex AI Pipeline
"Explain why model rejected application"Vertex AI Explainability (SHAP)
"Train on distributed hospital data"Federated Learning
"Prevent BigQuery data exfiltration"VPC Service Controls
"Compare model performance across runs"Vertex AI Experiments

Exam tip: The GCP Professional ML Engineer exam typically asks about architecture decisions, not API syntax. Key question patterns: "which service BEST fits the requirement", "what is the FIRST step", "which approach requires the LEAST operational overhead". Always prioritize GCP managed services when the question mentions "minimal management" or "serverless".

5. Study Plan

DayFocus
Day 1Vertex AI full platform: Training, Pipelines, Endpoints, Monitoring
Day 2Data engineering: Pub/Sub, Dataflow, Dataproc, Cloud Composer
Day 3BigQuery ML + Feature Engineering + Feature Store
Day 4Responsible AI: Explainability, Fairness, Privacy, Security
Day 5Practice exam 1 — identify weak areas
Day 6Review weak areas + Practice exam 2
Day 7Cheat sheet review only