Google Cloud Certified Professional Machine Learning Engineer (PMLE) practice questions

Domain coverage 6 domains

• Architecting ML Solutions (~18%)
• Data Processing and Feature Engineering (~18%)
• Developing ML Models (~22%)
• Automating ML Pipelines — MLOps (~22%)
• Deploying and Serving Models (~12%)
• Monitoring and Optimization (~8%)

Sample questions 7 of 65 shown

1. Q1 Framing ML Problems

   You need to develop an image classification model by using a large dataset that contains labeled images in a Cloud Storage bucket. What should you do?

   1. Use Vertex AI Pipelines with the Kubeflow Pipelines SDK to create a pipeline that reads the images from Cloud Storage and trains the model.
   2. Use Vertex AI Pipelines with TensorFlow Extended (TFX) to create a pipeline that reads the images from Cloud Storage and trains the model.
   3. Import the labeled images as a managed dataset in Vertex AI and use AutoML to train the model.
   4. Convert the image dataset to a tabular format using Dataflow Load the data into BigQuery and use BigQuery ML to train the model.

   C. Vertex AI managed datasets with AutoML provide a no-code solution for training image classification models on labeled data. Vertex AI Pipelines (A/B) are for orchestrating ML workflows, not direct training from labeled datasets. Converting images to tabular format (D) loses spatial information critical for image tasks.
2. Q2 Framing ML Problems

   You need to deploy a scikit-leam classification model to production. The model must be able to serve requests 24/7, and you expect millions of requests per second to the production application from 8 am to 7 pm. You need to minimize the cost of deployment. What should you do?

   1. Deploy an online Vertex AI prediction endpoint. Set the max replica count to 1
   2. Deploy an online Vertex AI prediction endpoint. Set the max replica count to 100
   3. Deploy an online Vertex AI prediction endpoint with one GPU per replica. Set the max replica count to 1
   4. Deploy an online Vertex AI prediction endpoint with one GPU per replica. Set the max replica count to 100

   B. A max replica count of 100 allows Vertex AI to autoscale horizontally to handle millions of requests per second during peak hours (8am-7pm) while minimizing cost by scaling down overnight. A max of 1 replica (A/C) cannot handle peak traffic. GPUs (C/D) add unnecessary cost for scikit-learn models that don't benefit from GPU acceleration.
3. Q3 Framing ML Problems

   Your company stores a large number of audio files of phone calls made to your customer call center in an on-premises database. Each audio file is in wav format and is approximately 5 minutes long. You need to analyze these audio files for customer sentiment. You plan to use the Speech-to-Text API You want to use the most efficient approach. What should you do?

   1. 1. Upload the audio files to Cloud Storage
   2. 1. Upload the audio files to Cloud Storage.
   3. 1. Iterate over your local files in Python
   4. 1. Iterate over your local files in Python

   B. For long audio files (>1 minute), the Speech-to-Text API requires async recognition. The correct workflow is: upload to Cloud Storage, then call async recognition. Option A uses long_running_recognize incorrectly with local files. Options C/D use sync recognition which is limited to ~1 minute audio.
4. Q4 Framing ML Problems

   You work as an analyst at a large banking firm. You are developing a robust scalable ML pipeline to tram several regression and classification models. Your primary focus for the pipeline is model interpretability. You want to productionize the pipeline as quickly as possible. What should you do?

   1. Use Tabular Workflow for Wide & Deep through Vertex AI Pipelines to jointly train wide linear models and deep neural networks
   2. Use Google Kubernetes Engine to build a custom training pipeline for XGBoost-based models
   3. Use Tabular Workflow for TabNet through Vertex AI Pipelines to train attention-based models
   4. Use Cloud Composer to build the training pipelines for custom deep learning-based models

   C. Tabular Workflow for TabNet through Vertex AI Pipelines provides built-in attention-based interpretability, which aligns with the model interpretability requirement. Wide & Deep (A) focuses on memorization vs generalization. XGBoost on GKE (B) requires more setup. Cloud Composer (D) is for workflow orchestration, not ML-specific interpretability.
5. Q5 Framing ML Problems

   You work for a hotel and have a dataset that contains customers’ written comments scanned from paper-based customer feedback forms, which are stored as PDF files. Every form has the same layout. You need to quickly predict an overall satisfaction score from the customer comments on each form. How should you accomplish this task?

   1. Use the Vision API to parse the text from each PDF file. Use the Natural Language API analyzeSentiment feature to infer overall satisfaction scores.
   2. Use the Vision API to parse the text from each PDF file. Use the Natural Language API analyzeEntitySentiment feature to infer overall satisfaction scores.
   3. Uptrain a Document AI custom extractor to parse the text in the comments section of each PDF file. Use the Natural Language API analyzeSentiment feature to infer overall satisfaction scores.
   4. Uptrain a Document AI custom extractor to parse the text in the comments section of each PDF file. Use the Natural Language API analyzeEntitySentiment feature to infer overall satisfaction scores.

   C. Document AI custom extractor can be trained to extract specific fields (like comments) from PDF forms with consistent layouts, then analyzeSentiment provides overall sentiment. Option A/B use Vision API which doesn't handle form-structured text extraction. analyzeEntitySentiment (B/D) focuses on entity-level, not overall sentiment.
6. Q6 Framing ML Problems

   You work for a startup that has multiple data science workloads. Your compute infrastructure is currently on-premises, and the data science workloads are native to PySpark. Your team plans to migrate their data science workloads to Google Cloud. You need to build a proof of concept to migrate one data science job to Google Cloud. You want to propose a migration process that requires minimal cost and effort. What should you do first?

   1. Create a n2-standard-4 VM instance and install Java, Scala, and Apache Spark dependencies on it.
   2. Create a Google Kubernetes Engine cluster with a basic node pool configuration, install Java, Scala, and Apache Spark dependencies on it.
   3. Create a Standard (1 master, 3 workers) Dataproc cluster, and run a Vertex AI Workbench notebook instance on it.
   4. Create a Vertex AI Workbench notebook with instance type n2-standard-4.

   C. Dataproc is the managed Spark/Hadoop service on GCP, requiring minimal migration effort for PySpark workloads. Creating VMs manually (A) or GKE clusters (B) requires more configuration. Vertex AI Workbench (D) alone doesn't run PySpark jobs.
7. Q7 Framing ML Problems

   You are analyzing customer data for a healthcare organization that is stored in Cloud Storage. The data contains personally identifiable information (PII). You need to perform data exploration and preprocessing while ensuring the security and privacy of sensitive fields. What should you do?

   1. Use the Cloud Data Loss Prevention (DLP) API to de-identify the PII before performing data exploration and preprocessing.
   2. Use customer-managed encryption keys (CMEK) to encrypt the PII data at rest, and decrypt the PII data during data exploration and preprocessing.
   3. Use a VM inside a VPC Service Controls security perimeter to perform data exploration and preprocessing.
   4. Use Google-managed encryption keys to encrypt the PII data at rest, and decrypt the PII data during data exploration and preprocessing.

   A. Cloud DLP API can automatically identify and de-identify PII before any data processing, ensuring security during exploration. CMEK (B) encrypts data but doesn't prevent PII exposure during processing. VPC SC (C) controls network access but doesn't de-identify data.

Answer all 65 questions with instant feedback →

Reader reviews 140 ratings

• Highly recommended for Google certification. The PMLE practice questions are worth every penny.

  ★★★★★★ Andrew M.

• The way the PMLE questions are broken down by domain is great. Let me focus on my weak spots without wasting time.

  ★★★★★★ Olivia H.

• Great resource for PMLE. I liked that I could jump straight to specific domains instead of going through everything in order.

  ★★★★★★ Dominic S.

• I recommend this to everyone preparing for PMLE. The lifetime access is great — I keep coming back for reference.

  ★★★★★★ Robert C.

• The PMLE practice exam was crucial to my success. The domains map perfectly to the official exam blueprint.

  ★★★★★★ Jennifer F.

• I’ve purchased a few Google exam dumps over the years and this PMLE one is easily the best quality.

  ★★★★★★ Samuel T.

Frequently asked questions

Free study resources

Community-verified analysis of 169 questions from real test-taker discussions — 93 deep analyses and 25 FAQs.

More Google exams