Google ML Engineer Beginn - Free Practice Quiz

Question 1 of 60

On the Google Professional ML Engineer 2023 exam, which Google Cloud service is the PRIMARY managed platform for training and deploying ML models?

A BigQuery ML B Cloud Dataflow C Cloud Run D Vertex AI

1 / 60

Question 2 of 60

On the Google Professional ML Engineer 2023 exam, which component of the ML development lifecycle involves checking model performance on held-out data before production deployment?

A Data ingestion B Feature engineering C Continuous training D Model evaluation

2 / 60

Question 3 of 60

On the Google Professional ML Engineer 2023 exam, which Vertex AI managed service provides pre-built, pre-trained ML APIs for vision, natural language, and translation tasks?

A Vertex AI Pipelines B Vertex AI AutoML C Vertex AI Custom Training D Vertex AI pre-built APIs (Cloud Vision API, Natural Language API, Translation API)

3 / 60

Question 4 of 60

On the Google Professional ML Engineer 2023 exam, which data format is MOST commonly used for storing large structured datasets for ML training in Google Cloud?

A Plain text .txt files B XML files in Cloud Bigtable C JSON documents in Firestore D CSV files stored in Cloud Storage or Avro/TFRecord formats for efficient ML consumption

4 / 60

Question 5 of 60

According to the Google Professional ML Engineer 2023 exam guide, which type of ML problem involves predicting a continuous numerical value?

A Classification B Clustering C Recommendation D Regression

5 / 60

Question 6 of 60

The Google Professional ML Engineer 2023 exam identifies which service as the managed notebook environment for ML experimentation on Google Cloud?

A Cloud Shell B Google Colab Enterprise C Cloud Dataproc D Vertex AI Workbench

6 / 60

Question 7 of 60

The Google Professional ML Engineer 2023 exam identifies which concept as training a model using multiple machines or GPUs simultaneously to reduce training time?

A Batch normalization B Transfer learning C Feature engineering D Distributed training

7 / 60

Question 8 of 60

The Google Professional ML Engineer 2023 exam describes which term as the process of selecting the most relevant input variables for an ML model to reduce complexity and improve generalization?

A Data augmentation B Model evaluation C Hyperparameter tuning D Feature selection

8 / 60

Question 9 of 60

The Google Professional ML Engineer 2023 exam covers 'feature engineering' as the process of:

A Splitting data into training and test sets B Tuning model hyperparameters C Deploying a model to production D Transforming raw data into informative inputs that improve ML model performance

9 / 60

Question 10 of 60

According to the Google Professional ML Engineer 2023 exam, a neural network 'epoch' refers to:

A A single forward pass through one training sample B The total number of layers in a neural network C A single mini-batch gradient update D One complete pass through the entire training dataset during training

10 / 60

Question 11 of 60

According to the Google Professional ML Engineer 2023 exam, which approach allows ML practitioners to automatically select the best model architecture and hyperparameters without manual tuning?

A Cloud Dataflow B Custom Training C BigQuery ML D AutoML

11 / 60

Question 12 of 60

According to the Google Professional ML Engineer 2023 exam, which Google Cloud service provides a fully managed message queue for streaming ML inference requests to a model serving endpoint?

A Cloud Spanner B BigQuery C Cloud Storage D Pub/Sub

12 / 60

Question 13 of 60

Which Google Cloud service, highlighted in the Google Professional ML Engineer 2023 exam, is used for large-scale, serverless data processing pipelines that feed ML training jobs?

A Cloud Storage B BigQuery C Cloud Composer D Cloud Dataflow

13 / 60

Question 14 of 60

On the Google Professional ML Engineer 2023 exam, which data split strategy is MOST appropriate when temporal order matters (e.g., time-series forecasting)?

A Random shuffle and split B K-fold cross-validation C Stratified random sampling D Time-based split (chronological train/validation/test)

14 / 60

Question 15 of 60

The Google Professional ML Engineer 2023 exam notes that 'precision' in a classification model measures:

A The fraction of all actual positive instances that the model correctly identified B The model's ability to generalize to unseen data C The model's overall accuracy across all classes D The fraction of predicted positive instances that are actually positive

15 / 60

Question 16 of 60

On the Google Professional ML Engineer 2023 exam, 'underfitting' occurs when a model:

A Achieves perfect accuracy on the training set B Memorizes training data but fails on new data C Is too simple to capture the underlying patterns in the data, resulting in high training and test error D Has too many layers for the available GPU memory

16 / 60

Question 17 of 60

On the Google Professional ML Engineer 2023 exam, which term describes a model that performs well on training data but poorly on unseen test data?

A Underfitting B Bias C Overfitting D Regularization

17 / 60

Question 18 of 60

The Google Professional ML Engineer 2023 exam covers which Google Cloud service for orchestrating end-to-end ML pipelines using containerized components?

A Cloud Composer B Cloud Build C Vertex AI Pipelines D Cloud Scheduler

18 / 60

Question 19 of 60

On the Google Professional ML Engineer 2023 exam, which Google Cloud hardware accelerator is SPECIFICALLY designed to accelerate TensorFlow ML workloads?

A Central Processing Unit (CPU) B Graphics Processing Unit (GPU) C Tensor Processing Unit (TPU) D Field-Programmable Gate Array (FPGA)

19 / 60

Question 20 of 60

The Google Professional ML Engineer 2023 exam identifies 'learning rate' as a hyperparameter that controls:

A The number of training samples processed per gradient update B The maximum number of training iterations C The step size by which model weights are updated in the direction of the gradient D The ratio of training data to validation data

20 / 60

Question 21 of 60

The Google Professional ML Engineer 2023 exam identifies which BigQuery capability as allowing SQL users to train and query ML models directly within BigQuery?

A BigQuery Data Transfer Service B BigQuery Omni C BigQuery ML D BigQuery BI Engine

21 / 60

Question 22 of 60

According to the Google Professional ML Engineer 2023 exam, 'model drift' occurs when:

A A model's weights exceed the memory capacity of the training hardware B A distributed training job fails partway through C The statistical distribution of real-world input data diverges from the training data distribution over time D A model's hyperparameters are set incorrectly before training

22 / 60

Question 23 of 60

The Google Professional ML Engineer 2023 exam describes 'A/B testing' for ML models in production as a method to:

A Compare a model's training accuracy against its validation accuracy B Evaluate model performance on different hardware configurations C Route a portion of live traffic to a new model variant and compare real-world metrics against the current production model D Test a model's performance on artificially generated adversarial examples

23 / 60

Question 24 of 60

According to the Google Professional ML Engineer 2023 exam, which technique is MOST effective for augmenting image training data to improve model robustness?

A Reducing image resolution to save storage space B Converting all images to grayscale before training C Applying random rotations, flips, crops, and color jitter to generate additional training variants D Using only images that are perfectly centered

24 / 60

Question 25 of 60

According to the Google Professional ML Engineer 2023 exam, which metric is MOST appropriate for evaluating a binary classification model on an imbalanced dataset?

A Accuracy B Mean Squared Error C AUC-ROC (Area Under the ROC Curve) D R-squared

25 / 60

Question 26 of 60

On the Google Professional ML Engineer 2023 exam, which loss function is MOST appropriate for a multi-class classification problem with mutually exclusive classes?

A Mean Squared Error B Binary Cross-Entropy C Categorical Cross-Entropy D Hinge Loss

26 / 60

Question 27 of 60

According to the Google Professional ML Engineer 2023 exam, which Vertex AI feature enables teams to track experiments, compare metrics, and reproduce training runs?

A Vertex AI Model Monitoring B Vertex AI Feature Store C Vertex AI Experiments D Vertex AI Vizier

27 / 60

Question 28 of 60

On the Google Professional ML Engineer 2023 exam, which Vertex AI managed service supports building recommendation systems using collaborative filtering?

A Vertex AI Feature Store B Vertex AI Model Monitoring C Vertex AI AutoML (Tabular) or custom training with collaborative filtering frameworks D Vertex AI Vizier

28 / 60

Question 29 of 60

The Google Professional ML Engineer 2023 exam notes that 'transfer learning' is MOST useful when:

A A large labeled dataset is available for the target task B The model must be trained from scratch for maximum accuracy C The source and target tasks share similar low-level features and labeled target data is scarce D Target data distribution is entirely different from the source domain

29 / 60

Question 30 of 60

The Google Professional ML Engineer 2023 exam describes 'batch prediction' (offline inference) as MOST appropriate when:

A Sub-millisecond response latency is required for each user request B Predictions are needed in real time for a user-facing application C Large volumes of predictions can be generated asynchronously without strict latency requirements D The model must continuously learn from new data streams

30 / 60

Question 31 of 60

The Google Professional ML Engineer 2023 exam identifies 'recall' in a classification model as:

A The fraction of actual positives correctly identified by the model B The fraction of predicted positives that are actually positive C The model's overall accuracy on the test set D The area under the precision-recall curve

31 / 60

Question 32 of 60

The Google Professional ML Engineer 2023 exam notes that 'gradient descent' minimizes a model's loss function by:

A Iteratively adjusting model parameters in the direction that reduces the loss B Randomly initializing weights at each training step C Removing the lowest-contributing features from the dataset D Applying L2 regularization to all model weights automatically

32 / 60

Question 33 of 60

On the Google Professional ML Engineer 2023 exam, which Vertex AI feature automates the process of finding optimal hyperparameters for ML model training?

A Vertex AI Vizier (Hyperparameter Tuning) B Vertex AI Model Monitoring C Vertex AI Pipelines D Vertex AI Feature Store

33 / 60

Question 34 of 60

According to the Google Professional ML Engineer 2023 exam, which activation function is most commonly used in hidden layers of deep neural networks because it mitigates the vanishing gradient problem?

A ReLU (Rectified Linear Unit) B Tanh C Sigmoid D Softmax

34 / 60

Question 35 of 60

On the Google Professional ML Engineer 2023 exam, which approach BEST addresses the problem of a model trained on biased or unrepresentative data?

A Auditing training data for representation issues, collecting additional data to correct imbalances, and applying fairness-aware training techniques B Increasing model complexity to compensate for data quality issues C Deploying the model and relying on feedback from users to identify bias D Reducing regularization to allow the model to fit the data more closely

35 / 60

Question 36 of 60

According to the Google Professional ML Engineer 2023 exam, which technique is used to serve an ML model efficiently on edge devices with limited compute resources?

A Model pruning and quantization to reduce model size and latency B Increasing model depth to improve accuracy before edge deployment C Migrating edge inference to cloud-based VMs D Training a separate, larger model exclusively for edge use

36 / 60

Question 37 of 60

According to the Google Professional ML Engineer 2023 exam, which ML problem type is BEST suited to a situation where you have no labeled data and want to discover natural groupings in data?

A Unsupervised clustering B Regression C Reinforcement learning D Supervised classification

37 / 60

Question 38 of 60

The Google Professional ML Engineer 2023 exam identifies which Vertex AI feature as monitoring deployed models for prediction quality degradation in production?

A Vertex AI Model Monitoring B Vertex AI Vizier C Vertex AI Feature Store D Vertex AI Workbench

38 / 60

Question 39 of 60

The Google Professional ML Engineer 2023 exam covers which technique for explaining individual model predictions to help users understand why a model produced a specific output?

A Explainable AI (XAI) â€” e.g., Shapley values, LIME, or Vertex AI Explainability B Model quantization to reduce precision C Pruning the model to reduce parameter count D Batch normalization applied to the output layer

39 / 60

Question 40 of 60

On the Google Professional ML Engineer 2023 exam, which concept describes encoding categorical variables as binary vectors where only one element is active?

A One-hot encoding B Embedding C Normalization D Tokenization

40 / 60

Question 41 of 60

The Google Professional ML Engineer 2023 exam identifies which Google Cloud storage service as the MOST appropriate for storing large ML training datasets?

A Cloud Storage (GCS) B Cloud Spanner C Cloud SQL D Cloud Bigtable

41 / 60

Question 42 of 60

On the Google Professional ML Engineer 2023 exam, which technique is MOST commonly used to handle class imbalance in a binary classification dataset?

A Oversampling the minority class (e.g., SMOTE) or using class-weighted loss functions B Reducing the learning rate C Increasing the number of training epochs D Using a more complex neural network architecture

42 / 60

Question 43 of 60

According to the Google Professional ML Engineer 2023 exam, which MLOps practice ensures that ML models are automatically retrained when performance degrades or data drift is detected?

A Continuous training (CT) â€” automated retraining triggered by monitoring alerts or scheduled pipelines B Manual retraining by data scientists after quarterly model reviews C Deploying a new model only when the old one fails completely D Using larger batch sizes during the initial training run

43 / 60

Question 44 of 60

The Google Professional ML Engineer 2023 exam identifies 'pipeline versioning' as important in MLOps because it:

A Enables reproducibility of ML workflows and rollback to previous pipeline versions when issues arise B Speeds up model training by caching intermediate outputs C Eliminates the need for container images in ML deployments D Automatically increases model accuracy with each new version

44 / 60

Question 45 of 60

On the Google Professional ML Engineer 2023 exam, 'data leakage' in ML refers to:

A Including information in training data that would not be available at prediction time, causing artificially high model performance B Network latency causing training data to arrive late C A model that leaks training data through its predictions D Data being lost during training due to hardware failure

45 / 60

Question 46 of 60

According to the Google Professional ML Engineer 2023 exam, which Google Cloud tool is used to explore, analyze, and visualize ML datasets to understand feature distributions and detect data quality issues?

A Cloud Data Catalog B TensorFlow Data Validation (TFDV) / Vertex AI Dataset C Cloud Composer D Cloud Spanner

46 / 60

Question 47 of 60

The Google Professional ML Engineer 2023 exam describes 'model quantization' as a technique used PRIMARILY to:

A Increase training accuracy by using higher precision floating point arithmetic B Reduce model size and inference latency by representing weights in lower precision (e.g., float32 â†’ int8) C Prevent overfitting by reducing the number of neurons in each layer D Distribute model training across multiple machines

47 / 60

Question 48 of 60

According to the Google Professional ML Engineer 2023 exam, which Google Cloud service is MOST appropriate for serving a TensorFlow SavedModel as a REST/gRPC endpoint with autoscaling?

A Cloud Functions B Vertex AI Online Prediction C Cloud Dataflow D Cloud Dataproc

48 / 60

Question 49 of 60

According to the Google Professional ML Engineer 2023 exam guide, which Vertex AI service provides a managed, centralized repository for storing, serving, and sharing ML features across teams?

A Vertex AI Workbench B Vertex AI Feature Store C Vertex AI Model Registry D Vertex AI Pipelines

49 / 60

Question 50 of 60

The Google Professional ML Engineer 2023 exam describes 'online serving' of ML models as MOST appropriate for:

A Generating quarterly reports from historical data B Serving real-time, low-latency predictions in response to individual inference requests C Training models on new data as it arrives D Processing petabyte-scale datasets overnight

50 / 60

Question 51 of 60

On the Google Professional ML Engineer 2023 exam, which Google Cloud service allows running ML workloads on distributed Apache Spark clusters?

A Cloud Dataflow B Cloud Dataproc C BigQuery D Cloud Composer

51 / 60

Question 52 of 60

The Google Professional ML Engineer 2023 exam describes 'batch size' in neural network training as:

A The total size of the training dataset in gigabytes B The number of training samples processed together in a single forward-backward pass before weights are updated C The number of model layers used during inference D The number of CPU cores allocated for distributed training

52 / 60

Question 53 of 60

The Google Professional ML Engineer 2023 exam describes the 'confusion matrix' as a tool PRIMARILY used to:

A Visualize the training loss curve over epochs B Evaluate the performance of a classification model by showing true vs. predicted labels C Compare hyperparameter tuning results D Measure model inference latency

53 / 60

Question 54 of 60

On the Google Professional ML Engineer 2023 exam, 'cross-validation' is used to:

A Merge multiple datasets into one training set B Estimate model performance more reliably by training and validating across multiple data folds C Deploy models across multiple Google Cloud regions D Compare the predictions of two different models simultaneously

54 / 60

Question 55 of 60

The Google Professional ML Engineer 2023 exam identifies which practice as packaging an ML model along with its runtime dependencies into a portable, reproducible unit for deployment?

A Saving model weights as a CSV file B Containerization (e.g., Docker) of the model serving environment C Exporting the model as a plain Python pickle file without dependencies D Storing the model in Cloud Bigtable

55 / 60

Question 56 of 60

On the Google Professional ML Engineer 2023 exam, which term describes the process of using model predictions as inputs for subsequent model training in an iterative feedback loop?

A Ensemble learning B Continuous training with feedback loops (e.g., active learning or online learning) C Transfer learning D Semi-supervised learning

56 / 60

Question 57 of 60

On the Google Professional ML Engineer 2023 exam, which technique is used to reduce model complexity and prevent overfitting by penalizing large model weights?

A Data augmentation B Regularization (L1 or L2) C Early stopping D Cross-validation

57 / 60

Question 58 of 60

According to the Google Professional ML Engineer 2023 exam, which preprocessing step is MOST important before training a regression model when features have very different numerical scales?

A One-hot encoding categorical features B Feature normalization or standardization (e.g., min-max scaling or z-score) C Removing all duplicate rows D Reducing the number of training epochs

58 / 60

Question 59 of 60

According to the Google Professional ML Engineer 2023 exam, which metric is MOST appropriate when optimizing a model for minimizing false negatives in a fraud detection system?

A Precision B Recall (sensitivity) C Accuracy D Specificity

59 / 60

Question 60 of 60

According to the Google Professional ML Engineer 2023 exam, which service provides managed workflows for building and deploying NLP models with pre-trained language model fine-tuning on Google Cloud?

A Cloud Dataflow B Vertex AI (with Generative AI support â€” fine-tuning foundation models) C Cloud Pub/Sub D Cloud Dataproc

60 / 60

Google ML Engineer Beginner Quiz