Google Google Professional Machine Learning Engineer Professional-Machine-Learning-Engineer Exam Dumps: Updated Questions & Answers (October 2025)

 The best option for configuring the workflow is to add the transformations as a preprocessing layer in the TensorFlow models. This option allows you to leverage the power and simplicity of TensorFlow to preprocess and transform the data with simple Python code. TensorFlow is a framework for building and training machine learning models. TensorFlow provides various tools and libraries for data analysis and machine learning. A preprocessing layer is a type of layer in TensorFlow that can perform data preprocessing and feature engineering operations on the input data. A preprocessing layer can help you customize the data transformation and preprocessing logic, and handle complex or non-standard data formats. A preprocessing layer can also help you minimize the preprocessing time, cost, and development effort, as you only need to write a few lines of code to implement the preprocessing layer, and you do not need to create any intermediate data sources or pipelines. By adding the transformations as a preprocessing layer in the TensorFlow models, you can use Vertex AI’s custom training service to train multiple TensorFlow models that read the data from BigQuery and predict future sales1.

The other options are not as good as option C, for the following reasons:

Option A: Writing the transformations into Spark that uses the spark-bigquery-connector and using Dataproc to preprocess the data would require more skills and steps than using a preprocessing layer in TensorFlow. Spark is a framework for distributed data processing and machine learning. Spark can read and write data from BigQuery by using the spark-bigquery-connector, which is a library that allows Spark to communicate with BigQuery. Dataproc is a service that can create and manage Spark clusters on Google Cloud. Dataproc can help you run Spark jobs on Google Cloud, and scale the clusters according to the workload. However, writing the transformations into Spark that uses the spark-bigquery-connector and using Dataproc to preprocess the data would require more skills and steps than using a preprocessing layer in TensorFlow. You would need to write code, create and configure the Spark cluster, install and import the spark-bigquery-connector, load and preprocess the data, and write the data back to BigQuery. Moreover, this option would create an intermediate data source in BigQuery, which can increase the storage and computation costs2.

Option B: Writing SQL queries to transform the data in-place in BigQuery would not allow you to use Vertex AI’s custom training service to train multiple TensorFlow models that read the data from BigQuery and predict future sales. BigQuery is a service that can perform data analysis and machine learning by using SQL queries. BigQuery can perform data transformation and preprocessing by using SQL functions and clauses, such as MIN, MAX, CASE, and TRANSFORM. BigQuery can also perform machine learning by using BigQuery ML, which is a feature that can create and train machine learning models by using SQL queries. However, writing SQL queries to transform the data in-place in BigQuery would not allow you to use Vertex AI’s custom training service to train multiple TensorFlow models that read the data from BigQuery and predict future sales. Vertex AI’s custom training service is a service that can run your custom machine learning code on Vertex AI. Vertex AI’s custom training service can support various machine learning frameworks, such as TensorFlow, PyTorch, and scikit-learn. Vertex AI’s custom training service cannot support SQL queries, as SQL is not a machine learning framework. Therefore, if you want to use Vertex AI’s custom training service, you cannot use SQL queries to transform the data in-place in BigQuery3.

Option D: Creating a Dataflow pipeline that uses the BigQueryIO connector to ingest the data, process it, and write it back to BigQuery would require more skills and steps than using a preprocessing layer in TensorFlow. Dataflow is a service that can create and run data processing and machine learning pipelines on Google Cloud. Dataflow can read and write data from BigQuery by using the BigQueryIO connector, which is a library that allows Dataflow to communicate with BigQuery. Dataflow can perform data transformation and preprocessing by using Apache Beam, which is a framework for distributed data processing and machine learning. However, creating a Dataflow pipeline that uses the BigQueryIO connector to ingest the data, process it, and write it back to BigQuery would require more skills and steps than using a preprocessing layer in TensorFlow. You would need to write code, create and configure the Dataflow pipeline, install and import the BigQueryIO connector, load and preprocess the data, and write the data back to BigQuery. Moreover, this option would create an intermediate data source in BigQuery, which can increase the storage and computation costs4.

References:

Preparing for Google Cloud Certification: Machine Learning Engineer, Course 3: Production ML Systems, Week 2: Serving ML Predictions

Google Cloud Professional Machine Learning Engineer Exam Guide, Section 2: Developing ML models, 2.1 Developing ML models by using TensorFlow

Official Google Cloud Certified Professional Machine Learning Engineer Study Guide, Chapter 4: Developing ML Models, Section 4.1: Developing ML Models by Using TensorFlow

TensorFlow Preprocessing Layers

Spark and BigQuery

Dataproc

BigQuery ML

Dataflow and BigQuery

Apache Beam

Model retraining is the process of updating an existing machine learning model with new data and parameters to improve its performance and accuracy. Model retraining is essential for maintaining the relevance and validity of the model, especially when the data or the environment changes over time. Model retraining can help to avoid or reduce the effects of model degradation, which is the phenomenon of the model’s predictive performance decreasing as it is tested on new datasets within rapidly evolving environments1.

For the use case of predicting sales numbers, model accuracy is crucial, because the production model is required to keep up with market changes. Market changes can affect the demand, supply, price, and preference of the products, and thus influence the sales numbers. If the model is not retrained with new data that reflects the market changes, it may become outdated and inaccurate, and fail to capture the patterns and trends of the sales numbers. Therefore, the most likely issue that is causing the steady decline in model accuracy is the lack of model retraining.

The other options are not as likely as option B, because they are not directly related to the model’s ability to adapt to market changes. Option A, poor data quality, may affect the model’s accuracy, but it is not a specific cause of model degradation over time. Option C, too few layers in the model for capturing information, may affect the model’s complexity and expressiveness, but it is not a specific cause of model degradation over time. Option D, incorrect data split ratio during model training, evaluation, validation, and test, may affect the model’s generalization and validation, but it is not a specific cause of model degradation over time. Therefore, option B, lack of model retraining, is the best answer for this question.

References:

Beware Steep Decline: Understanding Model Degradation In Machine Learning Models

Reasoning: The question asks for a design that serves asynchronous predictions to determine whether a machine part will fail. This means that the predictions do not need to be returned immediately to the sensors, but can be processed in batches and sent to a downstream system for monitoring. Option B is the only one that uses a streaming data pipeline with Pub/Sub and Dataflow, which can handle real-time data ingestion, processing, and prediction. Option B also invokes the model for prediction, which is required by the question. The other options either use synchronous predictions (option A), batch predictions (options C and D), or do not invoke the model for prediction (option D).

References: You can learn more about the differences between synchronous, asynchronous, and batch predictions in Vertex AI from this document. You can also find examples of how to use Pub/Sub and Dataflow for streaming data pipelines from this tutorial and this codelab.

According to the official exam guide1, one of the skills assessed in the exam is to “configure and optimize model monitoring jobs”. The Vertex AI Model Monitoring documentation states that “to reduce the cost of model monitoring, you can configure the sample rate of the requests that are logged and analyzed by model monitoring”. Therefore, decreasing the sample_rate parameter in the Randomsampleconfig of the monitoring job would reduce the model monitoring cost while continuing to quickly detect drift. The other options are not relevant or optimal for this scenario. References:

Preparing for Google Cloud Certification: Machine Learning Engineer Professional Certificate

Professional ML Engineer Exam Guide

Google Professional Machine Learning Certification Exam 2023

Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

[Vertex AI Model Monitoring]

Cloud Run can be triggered on new data arrivals, which makes it ideal for near-real-time processing. The function then initiates the Vertex AI Pipeline for preprocessing and storing features in Vertex AI Feature Store, aligning with the retraining needs. Cloud Scheduler (Option A) is suitable for scheduled jobs, not event-driven triggers. Dataflow (Option C) is better suited for batch processing or ETL rather than ML preprocessing pipelines.

=================

 The tf.data dataset is a TensorFlow API that provides a way to create and manipulate data pipelines for machine learning. The tf.data dataset allows you to apply various transformations to the data, such as reading, shuffling, batching, prefetching, and interleaving. These transformations can affect the performance and efficiency of the model training process1

One of the common performance issues in model training is input-bound, which means that the model is waiting for the input data to be ready and is not fully utilizing the computational resources. Input-bound can be caused by slow data loading, insufficient parallelism, or large data size. Input-bound can be detected by using the Cloud TPU profiler plugin, which is a tool that helps you analyze the performance of your model on Cloud TPUs. The Cloud TPU profiler plugin can show you the percentage of time that the TPU cores are idle, which indicates input-bound2

To reduce the input-bound bottleneck and speed up the model training process, you can make some modifications to the tf.data dataset. Two of the modifications that can help are:

Use the interleave option for reading data. The interleave option allows you to read data from multiple files in parallel and interleave their records. This can improve the data loading speed and reduce the idle time of the TPU cores. The interleave option can be applied by using the tf.data.Dataset.interleave method, which takes a function that returns a dataset for each input element, and a number of parallel calls3

Set the prefetch option equal to the training batch size. The prefetch option allows you to prefetch the next batch of data while the current batch is being processed by the model. This can reduce the latency between batches and improve the throughput of the model training. The prefetch option can be applied by using the tf.data.Dataset.prefetch method, which takes a buffer size argument. The buffer size should be equal to the training batch size, which is the number of examples per batch4

The other options are not effective or counterproductive. Reducing the value of the repeat parameter will reduce the number of epochs, which is the number of times the model sees the entire dataset. This can affect the model’s accuracy and convergence. Increasing the buffer size for the shuffle option will increase the randomness of the data, but also increase the memory usage and the data loading time. Decreasing the batch size argument in your transformation will reduce the number of examples per batch, which can affect the model’s stability and performance.

References: 1: tf.data: Build TensorFlow input pipelines 2: Cloud TPU Tools in TensorBoard 3: tf.data.Dataset.interleave 4: tf.data.Dataset.prefetch : [Better performance with the tf.data API]

 Oversampling is a technique for dealing with imbalanced datasets, where the majority class dominates the minority class. It balances the distribution of classes by increasing the number of samples in the minority class. Oversampling can improve the performance of a classifier by reducing the bias towards the majority class and increasing the sensitivity to the minority class.

In this case, the dataset includes transactions, of which 1% are identified as fraudulent. This means that the fraudulent transactions are the minority class and the non-fraudulent transactions are the majority class. A random forest model trained on this dataset might have a low recall for the fraudulent transactions, meaning that it might miss many of them and fail to detect fraud. This could have a high cost for the bank and its customers.

One way to overcome this problem is to oversample the fraudulent transactions 10 times, meaning that each fraudulent transaction is duplicated 10 times in the training dataset. This would increase the proportion of fraudulent transactions from 1% to about 10%, making the dataset more balanced. This would also make the random forest model more aware of the patterns and features that distinguish fraudulent transactions from non-fraudulent ones, and thus improve its accuracy and recall for the minority class.

For more information about oversampling and other techniques for imbalanced data, see the following references:

Random Oversampling and Undersampling for Imbalanced Classification

Exploring Oversampling Techniques for Imbalanced Datasets

According to the official exam guide1, one of the skills assessed in the exam is to “design, build, and productionalize ML models to solve business challenges using Google Cloud technologies”. Cloud Data Loss Prevention (DLP) API2 is a service that provides programmatic access to a powerful detection engine for personally identifiable information and other privacy-sensitive data in unstructured data streams, such as text blocks and images. Cloud DLP API helps you discover, classify, and protect your sensitive data by using techniques such as de-identification, masking, tokenization, and bucketing. You can use Cloud DLP API to de-identify the PII data before performing data exploration and preprocessing, and retain the data utility for ML purposes. Therefore, option A is the best way to perform data exploration and preprocessing while ensuring the security and privacy of sensitive fields. The other options are not relevant or optimal for this scenario. References:

Professional ML Engineer Exam Guide

Cloud Data Loss Prevention (DLP) API

Google Professional Machine Learning Certification Exam 2023

Latest Google Professional Machine Learning Engineer Actual Free Exam Questions

The best option to build a comprehensive system that recommends images to users that are similar in appearance to their own uploaded images is to download a pretrained convolutional neural network (CNN), and use the model to generate embeddings of the input images. Embeddings are low-dimensional representations of high-dimensional data that capture the essential features and semantics of the data. By using a pretrained CNN, you can leverage the knowledge learned from large-scale image datasets, such as ImageNet, and apply it to your own domain. A pretrained CNN can be used as a feature extractor, where the output of the last hidden layer (or any intermediate layer) is taken as the embedding vector for the input image. You can then measure the similarity between embeddings using a distance metric, such as cosine similarity or Euclidean distance, and recommend images that have the highest similarity scores to the user’s uploaded image. Option A is incorrect because downloading a pretrained CNN and fine-tuning the model to predict hashtags based on the input images may not capture the visual similarity of the images, as hashtags may not reflect the appearance of the images accurately. For example, two images of different breeds of dogs may have the same hashtag #dog, but they may not look similar to each other. Moreover, fine-tuning the model may require additional data and computational resources, and it may not generalize well to new images that have different or missing hashtags. Option B is incorrect because retrieving image labels and dominant colors from the input images using the Vision API may not capture the visual similarity of the images, as labels and colors may not reflect the fine-grained details of the images. For example, two images of the same breed of dog may have different labels and colors depending on the background, lighting, and angle of the image. Moreover, using the Vision API may incur additional costs and latency, and it may not be able to handle custom or domain-specific labels. Option C is incorrect because using the provided hashtags to create a collaborative filtering algorithm may not capture the visual similarity of the images, as collaborative filtering relies on the ratings or preferences of users, not the features of the images. For example, two images of different animals may have similar ratings or preferences from users, but they may not look similar to each other. Moreover, collaborative filtering may suffer from the cold start problem, where new images or users that have no ratings or preferences cannot be recommended. References:

Image similarity search with TensorFlow

Image embeddings documentation

Pretrained models documentation

Similarity metrics documentation