How should I use these Architecting low-code ML solutions practice questions?

Read each scenario carefully and choose your answer before revealing the explanation. Then check why your choice was right or wrong. Repeat until the reasoning feels automatic.

Can I practise just Architecting low-code ML solutions questions in a focused session?

Yes — use the session launcher on this page to start a 10-, 20-, 30- or 50-question session drawn entirely from the Architecting low-code ML solutions domain.

PMLE · topic practice

Architecting low-code ML solutions practice questions

Practise Google Professional Machine Learning Engineer Architecting low-code ML solutions practice questions — original exam-style scenarios with answer choices, explanations, and analysis of common mistakes.

Courseiva uses original exam-style practice questions designed for learning and revision. The goal is to understand the concepts, recognise exam patterns, and improve through explanations — not memorise copied exam dumps.

Reviewed byJohnson Ajibi· MSc IT Security

20 questionsDomain: Architecting low-code ML solutions

Practice 10 questions Browse domain →

What the exam tests

What to know about Architecting low-code ML solutions

Architecting low-code ML solutions questions test whether you can apply the concept in context, not just recognise a definition.

How the topic appears in realistic exam-style scenarios.

Which detail in the question changes the correct answer.

How to eliminate plausible but wrong options.

How to connect the question back to the wider exam objective.

Watch out for

Common Architecting low-code ML solutions exam traps

▸Answering from memory before reading the full scenario.
▸Missing a constraint such as cost, availability, security, scope or command context.
▸Choosing a broad answer when the question asks for the most specific fix.
▸Ignoring why the wrong options are tempting.

Practice set

Architecting low-code ML solutions questions

20 questions · select your answer, then reveal the explanation

Question 1mediummultiple choice

Read the full Architecting low-code ML solutions explanation →

A retail company wants to build a product recommendation system using BigQuery ML for their e-commerce platform. The data includes customer purchase history, product metadata, and clickstream logs. The ML engineer needs to minimize manual feature engineering and leverage pre-built solutions. Which approach should the engineer take?

Trap 1: Use a pre-built recommendation model from Vertex AI Model Garden…

This is not a BigQuery ML solution and requires additional infrastructure.

Trap 2: Write a custom TensorFlow model using the Vertex AI Training…

This requires significant custom coding, not low-code.

Trap 3: Export the data to CSV and use AutoML Tables to train a…

AutoML Tables is not integrated with BigQuery ML; exporting data adds complexity.

Study all Architecting low-code ML solutions common traps →

A
Use a pre-built recommendation model from Vertex AI Model Garden and deploy it to an endpoint.
Why wrong: This is not a BigQuery ML solution and requires additional infrastructure.
B
Write a custom TensorFlow model using the Vertex AI Training service and deploy it via Vertex AI Prediction.
Why wrong: This requires significant custom coding, not low-code.
C
Export the data to CSV and use AutoML Tables to train a recommendation model.
Why wrong: AutoML Tables is not integrated with BigQuery ML; exporting data adds complexity.
D
Use BigQuery ML's matrix factorization model (CREATE MODEL with model_type='matrix_factorization') to train directly on historical interaction data.
BigQuery ML provides low-code matrix factorization for recommendations.

Full breakdown with real-world context →

Question 2easymultiple choice

Read the full Architecting low-code ML solutions explanation →

A data scientist wants to quickly train a binary classification model on a tabular dataset stored in BigQuery without writing any code. They have limited ML experience. Which Google Cloud service should they use?

Trap 1: Vertex AI Workbench with a built-in scikit-learn notebook.

Requires Python coding and environment setup.

Trap 2: Dataflow with a TensorFlow pipeline.

Dataflow is for data processing, not training.

Trap 3: AutoML Tables with a direct BigQuery connection.

AutoML Tables is not SQL-based; it requires a UI or API.

Study all Architecting low-code ML solutions common traps →

A
Vertex AI Workbench with a built-in scikit-learn notebook.
Why wrong: Requires Python coding and environment setup.
B
Dataflow with a TensorFlow pipeline.
Why wrong: Dataflow is for data processing, not training.
C
BigQuery ML with CREATE MODEL statement using SQL.
BigQuery ML enables model creation with SQL, no coding required.
D
AutoML Tables with a direct BigQuery connection.
Why wrong: AutoML Tables is not SQL-based; it requires a UI or API.

Full breakdown with real-world context →

Question 3hardmultiple choice

Read the full Architecting low-code ML solutions explanation →

A company uses Vertex AI Pipelines to orchestrate their ML training workflow. The pipeline includes a BigQuery ML training step, a model evaluation step, and a deployment step to Vertex AI Endpoints. The engineer notices that the pipeline fails intermittently due to a quota exceeded error on Vertex AI Endpoints during model deployment. What is the best long-term solution to prevent this failure?

Trap 1: Run the pipeline steps sequentially with longer wait times.

May still hit quota if multiple pipelines run concurrently.

Trap 2: Switch to deploying models using a custom container on Compute…

Avoids Vertex AI endpoints but adds operational overhead.

Trap 3: Request a permanent quota increase for Vertex AI Endpoints.

Does not handle transient spikes and may not be approved quickly.

Study all Architecting low-code ML solutions common traps →

A
Run the pipeline steps sequentially with longer wait times.
Why wrong: May still hit quota if multiple pipelines run concurrently.
B
Add retry logic with exponential backoff to the deployment step in the pipeline.
Handles transient quota errors gracefully without manual intervention.
C
Switch to deploying models using a custom container on Compute Engine.
Why wrong: Avoids Vertex AI endpoints but adds operational overhead.
D
Request a permanent quota increase for Vertex AI Endpoints.
Why wrong: Does not handle transient spikes and may not be approved quickly.

Full breakdown with real-world context →

Question 4mediummultiple choice

Read the full Architecting low-code ML solutions explanation →

A manufacturing company wants to predict equipment failure using sensor data stored in BigQuery. They have limited ML expertise and want to use AutoML Tables. The data includes timestamps, numerical sensor readings, and a boolean 'failure' column. The dataset is highly imbalanced with only 1% failure cases. Which of the following is the most effective approach to handle the imbalance in AutoML Tables?

Trap 1: Downsample the majority class to balance the dataset.

Loses data and may reduce model accuracy.

Trap 2: Use a custom loss function in the training configuration.

AutoML Tables does not support custom loss functions.

Trap 3: Oversample the minority class using SQL before training.

AutoML Tables expects raw data; manual resampling may interfere with its optimizations.

Study all Architecting low-code ML solutions common traps →

A
Let AutoML Tables handle the imbalance automatically; it has built-in techniques for class imbalance.
AutoML Tables automatically adjusts for imbalance.
B
Downsample the majority class to balance the dataset.
Why wrong: Loses data and may reduce model accuracy.
C
Use a custom loss function in the training configuration.
Why wrong: AutoML Tables does not support custom loss functions.
D
Oversample the minority class using SQL before training.
Why wrong: AutoML Tables expects raw data; manual resampling may interfere with its optimizations.

Full breakdown with real-world context →

Question 5easymultiple choice

Read the full NAT/PAT explanation →

A marketing team wants to use a pre-built natural language processing (NLP) model from Vertex AI Model Garden to analyze customer feedback. They need to extract sentiment from text data stored in Cloud Storage. The team has no experience with model serving infrastructure. Which deployment option minimizes operational overhead?

Trap 1: Deploy the model as a Cloud Function invoked by Cloud Storage…

Not designed for low-latency inference with autoscaling.

Trap 2: Deploy the model as a Cloud Run service using a custom Docker…

Requires containerization and more configuration.

Trap 3: Deploy the model on App Engine flexible environment.

More complex and not optimized for ML inference.

Study all Architecting low-code ML solutions common traps →

A
Deploy the model as a Cloud Function invoked by Cloud Storage events.
Why wrong: Not designed for low-latency inference with autoscaling.
B
Deploy the model as a Cloud Run service using a custom Docker container.
Why wrong: Requires containerization and more configuration.
C
Deploy the model on App Engine flexible environment.
Why wrong: More complex and not optimized for ML inference.
D
Deploy the model to a Vertex AI Endpoint directly from Model Garden.
Simplest deployment with managed infrastructure.

Full breakdown with real-world context →

Question 6hardmultiple choice

Read the full Architecting low-code ML solutions explanation →

A financial institution uses BigQuery ML to train a linear regression model to predict loan default risk. The model is trained on a dataset with 100 million rows and 50 features. During inference, the engineer uses the ML.PREDICT function. However, the query takes several minutes to run and times out frequently. The data is static and updated monthly. What is the most cost-effective and low-code solution to improve prediction latency?

Trap 1: Create a Dataflow pipeline to precompute predictions and store them…

Adds complexity and cost, not low-code.

Trap 2: Use a materialized view to precompute the prediction features.

Does not apply ML model, only pre-aggregates data.

Trap 3: Increase the BigQuery compute capacity by reserving more slots.

Increases cost without addressing the root cause of repeated computation.

Study all Architecting low-code ML solutions common traps →

A
Export the trained model as a SQL function using the EXPORT MODEL statement, then use it for predictions.
Exports model as a persistent function for faster inference.
B
Create a Dataflow pipeline to precompute predictions and store them in a separate table.
Why wrong: Adds complexity and cost, not low-code.
C
Use a materialized view to precompute the prediction features.
Why wrong: Does not apply ML model, only pre-aggregates data.
D
Increase the BigQuery compute capacity by reserving more slots.
Why wrong: Increases cost without addressing the root cause of repeated computation.

Full breakdown with real-world context →

Question 7mediummulti select

Read the full Architecting low-code ML solutions explanation →

A company wants to build a low-code ML pipeline using Vertex AI Pipelines and BigQuery ML. They need to train, evaluate, and deploy a model. Which TWO statements are correct about the integration between Vertex AI Pipelines and BigQuery ML? (Choose TWO.)

Trap 1: BigQuery ML models are automatically stored in Vertex AI Model…

Manual registration is required.

Trap 2: BigQuery ML supports hyperparameter tuning using the CREATE MODEL…

BigQuery ML does not natively support hyperparameter tuning.

Trap 3: Vertex AI Pipelines supports automatic retry of failed steps due to…

Retry logic must be explicitly implemented.

Study all Architecting low-code ML solutions common traps →

A
BigQuery ML models are automatically stored in Vertex AI Model Registry after training.
Why wrong: Manual registration is required.
B
BigQuery ML supports hyperparameter tuning using the CREATE MODEL statement.
Why wrong: BigQuery ML does not natively support hyperparameter tuning.
C
Vertex AI Pipelines supports automatic retry of failed steps due to transient errors.
Why wrong: Retry logic must be explicitly implemented.
D
Vertex AI Pipeline steps can include BigQuery ML training via the BigQueryQueryJob operator.
BigQuery ML training can be invoked as a SQL query step.
E
The trained BigQuery ML model can be registered in Vertex AI Model Registry and deployed to an endpoint.
Model can be exported and registered.

Full breakdown with real-world context →

Question 8hardmulti select

Read the full NAT/PAT explanation →

A healthcare company uses AutoML Tables to predict patient readmission risk. The dataset contains 500,000 rows and 200 features, including patient demographics, lab results, and medical history. The model accuracy is lower than expected. The engineer wants to improve performance using low-code techniques. Which THREE actions are most effective? (Choose THREE.)

Trap 1: Increase the training time budget to the maximum allowed.

May not improve accuracy if data quality is the issue.

Trap 2: Use a custom model architecture via AutoML Tables advanced options.

AutoML Tables does not allow custom architecture.

Study all Architecting low-code ML solutions common traps →

A
Increase the training time budget to the maximum allowed.
Why wrong: May not improve accuracy if data quality is the issue.
B
Remove highly correlated features using AutoML Tables' built-in feature importance analysis.
Reduces noise and improves model generalization.
C
Engineer new features such as time since last admission and number of previous admissions.
Temporal features are often predictive in healthcare.
D
Use a custom model architecture via AutoML Tables advanced options.
Why wrong: AutoML Tables does not allow custom architecture.
E
Enable automated handling of missing values and outliers in the dataset configuration.
Improves data quality and model robustness.

Full breakdown with real-world context →

Question 9hardmultiple choice

Read the full Architecting low-code ML solutions explanation →

Refer to the exhibit. A data scientist trained a BigQuery ML classification model to detect fraudulent transactions. The dataset has 95% non-fraud (class 0) and 5% fraud (class 1). The evaluation metrics show high accuracy (0.91) but low recall (0.60) for fraud detection. Which low-code approach should the data scientist take to improve recall without significantly sacrificing precision?

Network Topology

Trap 1: Apply feature selection to reduce the number of features and focus…

May not improve recall and could reduce model performance.

Trap 2: Increase the number of training iterations by setting the…

Does not address the recall issue directly.

Trap 3: Re-train the model using AutoML Tables with class weights to…

AutoML Tables is not BigQuery ML; the model is already in BigQuery ML.

Study all Architecting low-code ML solutions common traps →

A
Use the ML.PREDICT function with a lower classification threshold (e.g., 0.3 instead of 0.5) to capture more positive cases.
Lowering the threshold increases recall by classifying more instances as positive.
B
Apply feature selection to reduce the number of features and focus on the most predictive ones.
Why wrong: May not improve recall and could reduce model performance.
C
Increase the number of training iterations by setting the MAX_ITERATIONS option to a higher value.
Why wrong: Does not address the recall issue directly.
D
Re-train the model using AutoML Tables with class weights to penalize false negatives more heavily.
Why wrong: AutoML Tables is not BigQuery ML; the model is already in BigQuery ML.

Full breakdown with real-world context →

Question 10mediummultiple choice

Read the full Architecting low-code ML solutions explanation →

Refer to the exhibit. A data engineer is defining a Vertex AI Pipeline step to train a model. The pipeline fails with an error: "Failed to create vertex ai custom job: Invalid resource name." What is the most likely cause of the error?

Network Topology

Trap 1: The container image URI is incorrect; it should be from…

The URI is valid for Vertex AI training.

Trap 2: The output artifact schema is missing the 'type' property.

Schema is not required and does not cause resource name errors.

Trap 3: The machine type n1-standard-4 is not supported for Vertex AI…

It is a supported machine type.

Study all Architecting low-code ML solutions common traps →

A
The container image URI is incorrect; it should be from gcr.io/vertex-ai/training.
Why wrong: The URI is valid for Vertex AI training.
B
The output artifact schema is missing the 'type' property.
Why wrong: Schema is not required and does not cause resource name errors.
C
The training_data input should be a Vertex AI Dataset resource, not a simple string.
The input expects a dataset resource name, not a raw string.
D
The machine type n1-standard-4 is not supported for Vertex AI training.
Why wrong: It is a supported machine type.

Full breakdown with real-world context →

Question 11hardmultiple choice

Read the full Architecting low-code ML solutions explanation →

You are an ML engineer at a logistics company. The company uses a Vertex AI Pipeline with BigQuery ML to train a model that predicts delivery delays based on weather, traffic, and historical order data. The pipeline runs daily and includes steps: (1) data extraction from BigQuery, (2) feature engineering using Dataflow, (3) model training with BigQuery ML (logistic regression), (4) model evaluation, and (5) conditional deployment to a Vertex AI Endpoint if accuracy > 0.85. Recently, the pipeline has been failing at step 5 with the error: "Vertex AI Endpoint creation failed: Quota limit of 1 endpoint per region exceeded." The company has already created one endpoint in the same region for another model. The pipeline is configured to create a new endpoint each time a model is deployed. The engineer needs to fix this with minimal changes to the pipeline code. Which course of action should the engineer take?

Trap 1: Submit a quota increase request to Google Cloud for Vertex AI…

Not a code fix; may take days to approve.

Trap 2: Change the region in the pipeline configuration to a region with…

May not be feasible and adds cross-region latency.

Trap 3: Remove the accuracy threshold and deploy every model automatically…

Deploys low-quality models and wastes resources.

Study all Architecting low-code ML solutions common traps →

A
Submit a quota increase request to Google Cloud for Vertex AI Endpoints in the current region.
Why wrong: Not a code fix; may take days to approve.
B
Change the region in the pipeline configuration to a region with available endpoint quota.
Why wrong: May not be feasible and adds cross-region latency.
C
Remove the accuracy threshold and deploy every model automatically to a pre-created endpoint.
Why wrong: Deploys low-quality models and wastes resources.
D
Modify the deployment step to check if an endpoint already exists and, if so, deploy a new model version to the existing endpoint instead of creating a new one.
Reuses the existing endpoint, avoiding quota limits.

Full breakdown with real-world context →

Question 12mediummultiple choice

Read the full Architecting low-code ML solutions explanation →

A retail company wants to build a customer churn prediction model using BigQuery ML. The data is stored in BigQuery tables and includes customer demographics, purchase history, and support interactions. The data scientist wants to experiment with different model types quickly without moving data to another environment. Which approach should they use?

Trap 1: Use Cloud Composer to orchestrate a custom training pipeline on…

Why D is wrong: Cloud Composer is for orchestration, not low-code model building.

Trap 2: Use AI Platform Notebooks with pandas and scikit-learn.

Why B is wrong: This is code-based, not low-code.

Trap 3: Export the data to Cloud Storage and use Vertex AI AutoML Tables.

Why A is wrong: This requires data movement and longer setup, not low-code.

Study all Architecting low-code ML solutions common traps →

A
Use Cloud Composer to orchestrate a custom training pipeline on Vertex AI.
Why wrong: Why D is wrong: Cloud Composer is for orchestration, not low-code model building.
B
Use AI Platform Notebooks with pandas and scikit-learn.
Why wrong: Why B is wrong: This is code-based, not low-code.
C
Use BigQuery ML to create and evaluate models directly in BigQuery.
Why C is correct: BigQuery ML is a low-code solution that works directly on BigQuery data.
D
Export the data to Cloud Storage and use Vertex AI AutoML Tables.
Why wrong: Why A is wrong: This requires data movement and longer setup, not low-code.

Full breakdown with real-world context →

Question 13easymultiple choice

Read the full Architecting low-code ML solutions explanation →

A company wants to implement a document processing solution that extracts key information from invoices and receipts. They have limited ML expertise and want to use a pre-trained solution as much as possible. Which Google Cloud service should they use?

Trap 1: AutoML Natural Language with custom entity extraction.

Why C is wrong: Requires custom training and labeled data, not low-code.

Trap 2: Vertex AI Workbench with custom Python scripts.

Why D is wrong: This is code-based, not low-code.

Trap 3: Cloud Vision API with OCR.

Why A is wrong: Vision API does not provide structured extraction specific to invoices.

Study all Architecting low-code ML solutions common traps →

A
Document AI with a pre-trained invoice processor.
Why B is correct: Document AI offers specialized pre-trained processors for invoices.
B
AutoML Natural Language with custom entity extraction.
Why wrong: Why C is wrong: Requires custom training and labeled data, not low-code.
C
Vertex AI Workbench with custom Python scripts.
Why wrong: Why D is wrong: This is code-based, not low-code.
D
Cloud Vision API with OCR.
Why wrong: Why A is wrong: Vision API does not provide structured extraction specific to invoices.

Full breakdown with real-world context →

Question 14hardmulti select

Read the full Architecting low-code ML solutions explanation →

Which TWO are best practices for implementing a low-code ML solution using Vertex AI AutoML? (Choose 2)

Trap 1: Impute missing values manually before uploading the dataset.

Why B is wrong: AutoML handles missing values automatically.

Trap 2: Normalize numerical features to zero mean and unit variance.

Why C is wrong: AutoML automatically normalizes features.

Trap 3: Export the data and train a custom model with a different…

Why E is wrong: This defeats the purpose of low-code AutoML.

Study all Architecting low-code ML solutions common traps →

A
Use the AutoML recommended data split (train/validation/test) to avoid overfitting.
Why A is correct: AutoML optimizes split for best performance.
B
Impute missing values manually before uploading the dataset.
Why wrong: Why B is wrong: AutoML handles missing values automatically.
C
Normalize numerical features to zero mean and unit variance.
Why wrong: Why C is wrong: AutoML automatically normalizes features.
D
Enable automatic feature engineering by leaving feature columns as raw data.
Why D is correct: AutoML performs feature engineering automatically.
E
Export the data and train a custom model with a different architecture.
Why wrong: Why E is wrong: This defeats the purpose of low-code AutoML.

Full breakdown with real-world context →

Question 15hardmultiple choice

Read the full Architecting low-code ML solutions explanation →

Refer to the exhibit. A data analyst creates a BigQuery ML logistic regression model for churn prediction. The model evaluation shows high precision but low recall. Which change to the model creation would most likely improve recall?

Exhibit

Refer to the exhibit.

```
# BigQuery ML model creation
CREATE OR REPLACE MODEL `mydataset.churn_model`
OPTIONS
  ( model_type='LOGISTIC_REG',
    auto_class_weights=TRUE,
    input_label_cols=['churned'] )
AS
SELECT
  * EXCEPT(customer_id, churn_date)
FROM `mydataset.training_data`
WHERE churn_date IS NOT NULL;

# Evaluation query
SELECT * FROM ML.EVALUATE(MODEL `mydataset.churn_model`);

# Prediction query
SELECT * FROM ML.PREDICT(MODEL `mydataset.churn_model`,
  TABLE `mydataset.new_customers`);
```

Trap 1: Drop more columns to reduce overfitting.

Why D is wrong: Overfitting may not be the issue, and dropping columns could worsen recall.

Trap 2: Increase the training data by including customers without churn…

Why B is wrong: Including non-churned customers may not improve recall.

Trap 3: Change model_type to 'BOOSTED_TREE_CLASSIFIER'.

Why A is wrong: Model type change may not directly improve recall.

Study all Architecting low-code ML solutions common traps →

A
Drop more columns to reduce overfitting.
Why wrong: Why D is wrong: Overfitting may not be the issue, and dropping columns could worsen recall.
B
Increase the training data by including customers without churn dates.
Why wrong: Why B is wrong: Including non-churned customers may not improve recall.
C
Use ML.ADJUST_THRESHOLD to lower the classification threshold.
Why C is correct: Lowering threshold increases sensitivity, improving recall.
D
Change model_type to 'BOOSTED_TREE_CLASSIFIER'.
Why wrong: Why A is wrong: Model type change may not directly improve recall.

Full breakdown with real-world context →

Question 16mediummultiple choice

Read the full Architecting low-code ML solutions explanation →

A financial services company uses Vertex AI AutoML Tables to build a credit risk model. The dataset contains 500,000 rows and 50 features, including loan amount, credit score, debt-to-income ratio, and employment length. The target variable is binary: 'default' (1) or 'no default' (0). The data is highly imbalanced, with only 2% defaults. The data scientist trains a model with AutoML Tables using default settings. The evaluation metrics show an AUC of 0.85, but the confusion matrix reveals that the model predicts 'no default' for almost all cases, missing most defaults. The data scientist needs to improve the model's ability to identify defaults without significantly increasing false positives. They have limited time and cannot write custom code. What should they do?

Trap 1: Manually split the data into a stratified train/test set to ensure…

Why B is wrong: AutoML already does stratified splitting; this won't improve recall.

Trap 2: Train multiple models with different algorithms (e.g., XGBoost,…

Why C is wrong: This requires custom code, not low-code.

Trap 3: Under-sample the majority class to create a balanced dataset and…

Why D is wrong: Under-sampling may lose valuable data and is not recommended with AutoML.

Study all Architecting low-code ML solutions common traps →

A
Manually split the data into a stratified train/test set to ensure the same proportion of defaults in each.
Why wrong: Why B is wrong: AutoML already does stratified splitting; this won't improve recall.
B
Train multiple models with different algorithms (e.g., XGBoost, Random Forest) and blend them using a custom script.
Why wrong: Why C is wrong: This requires custom code, not low-code.
C
Enable 'Enable weighted evaluation' and set the optimization objective to 'Maximize recall at a specific recall@P%' with a target precision of 0.5.
Why A is correct: AutoML Tables supports custom optimization objectives to handle imbalance.
D
Under-sample the majority class to create a balanced dataset and retrain.
Why wrong: Why D is wrong: Under-sampling may lose valuable data and is not recommended with AutoML.

Full breakdown with real-world context →

Question 17mediummultiple choice

Read the full Architecting low-code ML solutions explanation →

A retail company wants to build a customer churn prediction model using AutoML Tables. They have a dataset with 5000 rows and 50 features, including customer ID, transaction history, and support tickets. The target is a binary column 'churned'. After training, the model shows high accuracy but low recall for the churned class. What is the most likely cause?

Trap 1: The dataset is too small for AutoML to train effectively.

5000 rows is sufficient for AutoML; the issue is class imbalance.

Trap 2: The features are not normalized, leading to biased predictions.

AutoML handles scaling internally.

Trap 3: The dataset includes a unique customer ID feature, causing…

Customer ID is not predictive; AutoML ignores high-cardinality features.

Study all Architecting low-code ML solutions common traps →

A
The dataset is too small for AutoML to train effectively.
Why wrong: 5000 rows is sufficient for AutoML; the issue is class imbalance.
B
The features are not normalized, leading to biased predictions.
Why wrong: AutoML handles scaling internally.
C
The churned class is underrepresented, causing the model to favor the majority class.
Class imbalance leads to high accuracy but low recall for minority class.
D
The dataset includes a unique customer ID feature, causing overfitting.
Why wrong: Customer ID is not predictive; AutoML ignores high-cardinality features.

Full breakdown with real-world context →

Question 18mediumdrag order

Read the full Architecting low-code ML solutions explanation →

Drag and drop the steps to set up data lineage tracking for ML pipelines using Vertex AI Experiments in the correct order.

Drag steps to the numbered slots on the right, or tap a step then tap a slot.

Steps

Order

1Step 1

2Step 2

3Step 3

4Step 4

5Step 5

Question 19mediummatching

Read the full Architecting low-code ML solutions explanation →

Match each model evaluation metric to its use case.

Drag a concept onto its matching description — or click a concept then click the description.

Concepts

Matches

Measure of false positives in classification

Measure of false negatives in classification

Harmonic mean of precision and recall

Root mean squared error for regression

Cross-entropy loss for probabilistic classification

Question 20easymultiple choice

Read the full Architecting low-code ML solutions explanation →

A marketing team wants to analyze customer reviews for sentiment without writing code. Which Google Cloud service should they use?

Trap 1: Cloud Dataflow

Wrong: For data processing pipelines, not sentiment analysis.

Trap 2: Vertex AI Workbench

Wrong: Is a code-based environment, not low-code.

Trap 3: BigQuery ML

Wrong: Requires SQL and is for tabular models, not pre-trained sentiment.

Study all Architecting low-code ML solutions common traps →

A
Cloud Dataflow
Why wrong: Wrong: For data processing pipelines, not sentiment analysis.
B
Vertex AI Workbench
Why wrong: Wrong: Is a code-based environment, not low-code.
C
BigQuery ML
Why wrong: Wrong: Requires SQL and is for tabular models, not pre-trained sentiment.
D
Cloud Natural Language API
Correct: Pre-trained, no-code sentiment analysis.

Full breakdown with real-world context →

Continue with 20-question session →

Free account

Track your progress over time

Create a free account to save your results and see which topics improve across sessions.

Focused Architecting low-code ML solutions sessions

Start a Architecting low-code ML solutions only practice session

Every question in these sessions is drawn from the Architecting low-code ML solutions domain — nothing else.

10 questions 20 questions 30 questions 50 questions

Browse all Architecting low-code ML solutions questions →Mixed PMLE session

Frequently asked questions

What does the PMLE exam test about Architecting low-code ML solutions?: Architecting low-code ML solutions questions test whether you can apply the concept in context, not just recognise a definition.
How should I use these practice questions?: Select your answer before revealing the explanation. Then read why each option is right or wrong — this active recall approach builds retention far faster than re-reading notes.
Can I practise just Architecting low-code ML solutions questions in a focused session?: Yes — the session launcher on this page draws every question from the Architecting low-code ML solutions domain. Use a 10-question session first to gauge your baseline, then move to 20 or 30 once the weak spots are clear.
Where can I practise other PMLE topics?: Use the topic links above to move to related areas, or go back to the PMLE question bank to see all topics.
Are these real exam questions or dumps?: These are original practice questions written to test the same concepts the PMLE exam covers. They are not copied from any real exam or dump site.

Architecting low-code ML solutions only

10 questions 20 questions 30 questions 50 questions

Mixed PMLE session

Track your progress

A free account saves results across sessions and highlights which topics need work.

Study resources

All PMLE questions Architecting low-code ML solutions domain overview PMLE exam guide

Exam traps to avoid

▸Answering from memory before reading the full scenario.
▸Missing a constraint such as cost, availability, security, scope or command context.
▸Choosing a broad answer when the question asks for the most specific fix.
▸Ignoring why the wrong options are tempting.

Architecting low-code ML solutions practice questions

What to know about Architecting low-code ML solutions

Common Architecting low-code ML solutions exam traps

Architecting low-code ML solutions questions

A data scientist wants to quickly train a binary classification model on a tabular dataset stored in BigQuery without writing any code. They have limited ML experience. Which Google Cloud service should they use?

A company wants to build a low-code ML pipeline using Vertex AI Pipelines and BigQuery ML. They need to train, evaluate, and deploy a model. Which TWO statements are correct about the integration between Vertex AI Pipelines and BigQuery ML? (Choose TWO.)

Refer to the exhibit. A data engineer is defining a Vertex AI Pipeline step to train a model. The pipeline fails with an error: "Failed to create vertex ai custom job: Invalid resource name." What is the most likely cause of the error?

A company wants to implement a document processing solution that extracts key information from invoices and receipts. They have limited ML expertise and want to use a pre-trained solution as much as possible. Which Google Cloud service should they use?

Which TWO are best practices for implementing a low-code ML solution using Vertex AI AutoML? (Choose 2)

Refer to the exhibit. A data analyst creates a BigQuery ML logistic regression model for churn prediction. The model evaluation shows high precision but low recall. Which change to the model creation would most likely improve recall?

Exhibit

Drag and drop the steps to set up data lineage tracking for ML pipelines using Vertex AI Experiments in the correct order.

Match each model evaluation metric to its use case.

A marketing team wants to analyze customer reviews for sentiment without writing code. Which Google Cloud service should they use?

Track your progress over time

Start a Architecting low-code ML solutions only practice session

Related PMLE topic practice pages

Scaling prototypes into ML models practice questions

Automating and orchestrating ML pipelines practice questions

Collaborating within and across teams to manage data and models practice questions