A retail company wants to build a product recommendation engine using their existing BigQuery sales data. They have limited ML expertise on their team. Which GCP approach best fits this scenario?
Architecting Low Code Ml Solutions
Question · Easy
Spot the AI hallucination
Short
Your team needs to quickly build a document classification system for internal support tickets. The dataset contains 10,000 labeled examples across 15 categories. Which Vertex AI feature provides the fastest path to a production-ready model?
Chapter 1: Exam Overview & the Agent Platform Map
1.1 PMLE Exam Facts, Policies & the June 2026 Blueprint
Study guide
A data science team stores features in multiple BigQuery tables and Cloud Storage buckets. Different models reuse the same features but compute them independently, leading to training-serving skew. What should you implement to ensure feature consistency?
Collaborating To Manage Data And Models
Question · Medium
Tech Exam Prep Podcast
Listen on Spotify
Podcast
You are designing an ML pipeline where multiple teams contribute datasets stored in different GCP projects. You need to track data lineage across these projects and ensure reproducibility. Which service should you use?
Your team is developing a fraud detection model. The dataset has 99.5% legitimate transactions and 0.5% fraudulent ones. Which technique should you prioritize to handle this class imbalance?
Scaling Prototypes Into Ml Models
You need to train a deep learning model on a dataset that does not fit into the memory of a single GPU. The model architecture itself fits in a single GPU's memory. What is the most appropriate distributed training strategy?
1.2 From Vertex AI to Gemini Enterprise Agent Platform: The Product Name Map
Which Vertex AI feature allows you to run multiple training experiments simultaneously, track hyperparameters and metrics, and compare results across runs?
A healthcare company wants to use a pre-trained foundation model to analyze medical images but needs to fine-tune it on their proprietary radiology dataset without sending data outside their VPC. Which approach should they use?
You are preparing a tabular dataset for ML training that contains both numerical and categorical features. The numerical features have vastly different scales, and the categorical features have high cardinality. Which preprocessing combination is most appropriate?
Your model is deployed on a Vertex AI endpoint and receives variable traffic throughout the day, from 10 requests per second at night to 1,000 requests per second during peak hours. How should you configure the endpoint to optimize cost and latency?
Serving And Scaling Models
Chapter 2: Low-Code ML with BigQuery ML
2.1 Choosing, Training & Predicting with BigQuery ML Models
After deploying a credit scoring model to production, you notice that the model's precision for loan approvals has dropped by 15% over three months. Input feature distributions appear unchanged. What is the most likely cause?
Monitoring Ai Solutions
Question · Hard
You need to deploy a TensorFlow model that requires GPU acceleration for inference with sub-100ms latency. The model receives consistent traffic of approximately 500 requests per second. Which Vertex AI deployment option is most appropriate?
You want to build a Vertex AI Pipeline that trains a model, evaluates it against a baseline, and conditionally deploys it only if the evaluation metrics exceed a threshold. Which component should control the conditional deployment logic?
Automating And Orchestrating Ml Pipelines
Your organization wants to implement ML model governance. Multiple teams deploy models to production, and you need a centralized view of all deployed models with version history, evaluation metrics, and deployment status. Which Vertex AI component addresses this?
2.2 Feature Engineering & Feature Selection in BigQuery ML
An e-commerce company wants to serve personalized recommendations with sub-10ms latency using pre-computed embedding vectors. They have 50 million product embeddings. Which GCP service is designed for this use case?
You are building a CI/CD pipeline for ML models. After a model is retrained, you want to automatically run validation tests, compare performance against the current production model, and deploy only if the new model is superior. Which GCP services should you use together?
You are using Vertex AI to train a custom model with a TensorFlow training script. You want to optimize hyperparameters across learning rate, batch size, and dropout rate. Which Vertex AI feature should you use?
A production ML model receives prediction requests that occasionally contain missing feature values. You need to handle these gracefully without failing the prediction. What is the recommended approach?
2.3 Fine-Tuning & Evaluating Gemini Models from BigQuery
You are monitoring a deployed sentiment analysis model and notice that the distribution of input text lengths has shifted significantly from the training data distribution. The model's accuracy has not yet declined. What action should you take?
You want to implement canary deployment for a new version of your ML model on a Vertex AI endpoint. How should you configure this?
Your Vertex AI Pipeline needs to preprocess a 10 TB dataset by applying complex transformations including windowed aggregations and joins. Which GCP service should handle this data processing step within the pipeline?
A model predicting customer churn uses feature attribution to explain individual predictions. The product team wants to understand which features most influence churn predictions globally. Which Vertex AI Explainable AI method provides global feature importance?
Chapter 3: AutoML & Google Cloud AI APIs
3.1 Training Models with Agent Platform AutoML
You need to schedule a Vertex AI Pipeline to retrain a model daily using fresh data from BigQuery. Which orchestration approach is recommended?
Your team is training a large language model and wants to reduce training costs on Vertex AI without significantly impacting training time. Which approach offers the best cost-performance tradeoff?
An ML engineer discovers that their image classification model performs well on the test set but poorly in production. Investigation reveals that the training pipeline resizes images to 224×224 with one aspect-ratio policy, while the production serving container uses a different resize/crop policy on mobile-camera images. What type of problem is this?
You are using Vertex AI to deploy a model that must comply with HIPAA regulations. Which combination of features ensures data protection?
3.2 Building with Document AI, Vision, Translation & Other AI APIs
You have a TensorFlow model that takes 45 minutes to generate predictions for 100 million records stored in BigQuery. The predictions are needed for a daily batch report. Which approach minimizes cost while meeting the daily schedule?
Your team built a generative AI application using a foundation model from Model Garden. Users report that the model sometimes generates factually incorrect information about your company's products. What is the most effective way to reduce these hallucinations?
You are evaluating a binary classification model and need to select a metric that is robust to class imbalance. The business stakeholder prioritizes minimizing false negatives. Which metric should you report?
You are building a Vertex AI Pipeline that must handle failures gracefully. A training step occasionally fails due to transient infrastructure issues. How should you configure retry behavior?
Chapter 4: Foundation Models in Model Garden
4.1 Evaluating & Selecting Models in Model Garden
A Vertex AI model monitoring job detects that prediction drift has exceeded the configured threshold. No data drift is observed. What is the most likely explanation?
You need to create a data labeling job for 50,000 images to train an object detection model. Some images contain sensitive content requiring domain expertise. Which Vertex AI approach should you use?
You are designing an ML system that must process streaming sensor data from IoT devices and make real-time anomaly predictions. The data arrives via Cloud Pub/Sub at 10,000 events per second. Which architecture is most appropriate?
Your organization wants to use Google's Gemini model via Vertex AI to build a customer support chatbot. You need to ensure the model only answers questions about your products and politely declines off-topic queries. What should you implement?
4.2 Building & Tuning Solutions with Gemini, Imagen, Veo & MaaS
You want to evaluate a generative AI model's responses for safety, factual grounding, and helpfulness. Which Vertex AI service provides automated evaluation of these dimensions?
A financial services company requires that all ML models meet fairness criteria before deployment. They need to check whether a loan approval model produces equitable outcomes across demographic groups. Which approach should they use?
You need to transform raw text data into numerical embeddings for a semantic search application. Which Vertex AI API provides pre-trained text embedding generation?
Loading more…