Why Study with PlanetCert?
The Latest Questions
Practice questions and exam topics aligned with the current exam objectives.
Detailed Explanations
Go beyond the answer. Master the material with comprehensive learning and professional explanations for every concept.
AI-Powered Insights
Personalized preparation guidance that adapts to your performance and identifies weak spots automatically.
Exam Information
Official specifications published by Google
Exam Format
120 min
50-60
70%
Professional
Registration
$200 USD
Kryterion Testing Centers or online proctoring
English, Japanese
Validity
2 years
Retake the exam within 60 days before certification expiration
GCP-PMLE Exam Topics and Domains
GCP-PMLE is organized into 6 weighted domains. Expect to work with Vertex AI, BigQuery, Vertex AI Prediction, Cloud Storage, and more.
1
Architecting Low-Code AI Solutions
13%
Developing ML models by using BigQuery ML
Building the appropriate BigQuery ML model based on the business problemFeature engineering or selection by using BigQuery MLGenerating predictions by using BigQuery ML
- Build appropriate BigQuery ML models (linear/binary classification, regression, time-series, matrix factorization, boosted trees, autoencoders) based on business problems
- Perform feature engineering or selection using BigQuery ML
- Generate predictions using BigQuery ML
Building AI solutions by using ML APIs or foundation models
Building applications by using ML APIs from Model GardenBuilding applications by using industry-specific APIsImplementing retrieval augmented generation (RAG) applications by using Vertex AI Agent Builder
- Build applications using ML APIs from Model Garden
- Build applications using industry-specific APIs (e.g., Document AI API, Retail API)
- Implement retrieval augmented generation (RAG) applications using Vertex AI Agent Builder
Training models by using AutoML
Preparing data for AutoMLUsing available data to train custom modelsUsing AutoML for tabular dataCreating forecasting models by using AutoMLConfiguring and debugging trained models
- Prepare data for AutoML (e.g., feature selection, data labeling, Tabular Workflows on AutoML)
- Use available data (tabular, text, speech, images, videos) to train custom models
- Use AutoML for tabular data
- Create forecasting models using AutoML
- Configure and debug trained models
2
Collaborating Within and Across Teams to Manage Data and Models
14%
Exploring and preprocessing organization-wide data
Organizing different types of data for efficient trainingManaging datasets in Vertex AIData preprocessingCreating and consolidating features in Vertex AI Feature StorePrivacy implications of data usage and/or collectionIngesting different data sources into Vertex AI for inference
- Organize different types of data (tabular, text, speech, images, videos) for efficient training
- Manage datasets in Vertex AI
- Perform data preprocessing using Dataflow, TFX, and BigQuery
- Create and consolidate features in Vertex AI Feature Store
- Understand privacy implications of data usage/collection (PII, PHI)
- Ingest different data sources into Vertex AI for inference
Model prototyping using Jupyter notebooks
Choosing the appropriate Jupyter backend on Google CloudApplying security best practices in Vertex AI WorkbenchUsing Spark kernelsIntegrating code source repositoriesDeveloping models in Vertex AI Workbench using common frameworksLeveraging foundation and open-source models in Model Garden
- Choose appropriate Jupyter backend on Google Cloud (Vertex AI Workbench, Colab Enterprise, Dataproc)
- Apply security best practices in Vertex AI Workbench
- Use Spark kernels for distributed processing
- Integrate code source repositories
- Develop models using common frameworks (TensorFlow, PyTorch, sklearn, Spark, JAX)
- Leverage foundation and open-source models in Model Garden
Tracking and running ML experiments
Choosing the appropriate Google Cloud environment for development and experimentationEvaluating generative AI solutions
- Choose the appropriate Google Cloud environment for development and experimentation given the framework
- Evaluate generative AI solutions effectively
3
Scaling Prototypes into ML Models
18%
Building models
Choosing ML framework and model architectureModeling techniques given interpretability requirements
- Choose appropriate ML framework and model architecture
- Apply modeling techniques that meet interpretability requirements
Training models
Organizing training data on Google CloudIngestion of various file types into trainingTraining using different SDKsUsing distributed training to organize reliable pipelinesHyperparameter tuningTroubleshooting ML model training failuresFine-tuning foundation models
- Organize training data on Google Cloud (Cloud Storage, BigQuery)
- Ingest various file types into training
- Train using different SDKs (Vertex AI custom training, Kubeflow on GKE, AutoML, tabular workflows)
- Use distributed training to organize reliable pipelines
- Perform hyperparameter tuning
- Troubleshoot ML model training failures
- Fine-tune foundation models using Vertex AI and Model Garden
Choosing appropriate hardware for training
Evaluation of compute and accelerator optionsDistributed training with TPUs and GPUs
- Evaluate compute and accelerator options (CPU, GPU, TPU, edge devices)
- Implement distributed training with TPUs and GPUs using Reduction Server on Vertex AI and Horovod
4
Serving and Scaling Models
20%
Serving models
Batch and online inferenceUsing different frameworks to serve modelsOrganizing a model registryA/B testing different versions of a model
- Implement batch and online inference using Vertex AI, Dataflow, BigQuery ML, and Dataproc
- Use different frameworks (PyTorch, XGBoost) to serve models
- Organize a model registry
- A/B test different versions of a model
Scaling online model serving
Vertex AI Feature Store for servingVertex AI public and private endpointsChoosing appropriate hardware for servingScaling the serving backend based on throughputTuning ML models for training and serving in production
- Use Vertex AI Feature Store for online serving
- Configure Vertex AI public and private endpoints
- Choose appropriate hardware for serving (CPU, GPU, TPU, edge)
- Scale serving backend based on throughput using Vertex AI Prediction and containerized serving
- Tune ML models for production (simplification, performance optimization, latency, memory, throughput)
5
Automating and Orchestrating ML Pipelines
22%
Developing end-to-end ML pipelines
Data and model validationEnsuring consistent data pre-processing between training and servingHosting third-party pipelines on Google CloudIdentifying components, parameters, triggers, and compute needsOrchestration framework selectionHybrid or multicloud strategiesSystem design with TFX components or Kubeflow DSL
- Implement data and model validation in pipelines
- Ensure consistent data pre-processing between training and serving
- Host third-party pipelines (e.g., MLFlow) on Google Cloud
- Identify components, parameters, triggers, and compute needs (Cloud Build, Cloud Run)
- Select appropriate orchestration framework (Kubeflow Pipelines, Vertex AI Pipelines, Cloud Composer)
- Design hybrid or multicloud ML strategies
- Design systems with TFX components or Kubeflow DSL
Automating model retraining
Determining an appropriate retraining policyCI/CD model deployment
- Determine appropriate retraining policies
- Implement CI/CD model deployment using Cloud Build and Jenkins
Tracking and auditing metadata
Tracking and comparing model artifacts and versionsHooking into model and dataset versioningModel and data lineage
- Track and compare model artifacts and versions using Vertex AI Experiments and Vertex ML Metadata
- Hook into model and dataset versioning
- Implement model and data lineage tracking
6
Monitoring AI Solutions
13%
Identifying risks to AI solutions
Building secure AI systemsAligning with Google's Responsible AI practicesAssessing AI solution readinessModel explainability on Vertex AI
- Build secure AI systems that protect against unintentional exploitation of data or models
- Align with Google's Responsible AI practices (monitoring for bias)
- Assess AI solution readiness (fairness, bias)
- Implement model explainability on Vertex AI
Monitoring, testing, and troubleshooting AI solutions
Establishing continuous evaluation metricsMonitoring for training-serving skewMonitoring for feature attribution driftMonitoring model performance against baselinesMonitoring for common training and serving errors
- Establish continuous evaluation metrics using Vertex AI Model Monitoring and Explainable AI
- Monitor for training-serving skew
- Monitor for feature attribution drift
- Monitor model performance against baselines, simpler models, and across time
- Monitor for common training and serving errors
How do I earn this certification?
Passing GCP-PMLE earns the Google Cloud Certified - Professional Machine Learning Engineer certification. It sits in the Machine Learning / AI track.
Current Level Exams
Next Level Options
- GCP-PDE - Professional Data EngineerDeepens data engineering skills that complement ML engineering
- GCP-PCA - Professional Cloud ArchitectBroadens architecture skills for designing complete cloud solutions including AI/ML
- GCP-PCDO - Professional Cloud DevOps EngineerExtends MLOps knowledge into broader DevOps and SRE practices
Alternative Paths
- GCP-PDE - Professional Data EngineerData engineering skills are foundational to ML engineering; often pursued together
- GCP-PDBE - Professional Cloud Database EngineerDatabase management skills support data infrastructure for ML workloads
- GCP-PNE - Professional Cloud Network EngineerNetwork engineering knowledge supports secure and performant ML serving architectures
- AWS-MLS - AWS Certified Machine Learning - Specialty AWS ML certification for multi-cloud ML engineering skills
- DP-100 - Microsoft Azure Data Scientist AssociateAzure ML certification for multi-cloud ML engineering skills
Practice with Precision
The PlanetCert Simulator mirrors the real exam environment with authentic questions and timed pressure.
How to study for this exam?
The most effective way to prepare for GCP-PMLE is by using the PlanetCert Simulator to practice questions and review detailed explanations.
What's changed on this exam?
Current Status
- ACTIVE
- Last content update: 2024-10-01
Updates
- Vertex AI Model Garden Current Now a core exam topic for building AI solutions with pre-trained and foundation models • Release date: 2024-10-01
- Vertex AI Agent Builder Current New exam topic for implementing RAG applications and enterprise search • Release date: 2024-10-01
- Gemini Models 2.0 Available through Model Garden; candidates should understand model selection and capabilities • Release date: 2025
- Vertex AI Feature Store 2.0 Updated Feature Store architecture with improved online/offline serving • Release date: 2024
Who should take this exam?
- 3+ years of industry experience including 1+ years designing and managing solutions using Google Cloud
- Proficiency in Python and SQL
- Familiarity with TensorFlow, PyTorch, or scikit-learn
- Hands-on experience with Vertex AI, BigQuery ML, and AutoML
- Understanding of distributed data processing tools