CloudExamCheap

PMLE Practice Exam: Professional Machine Learning Engineer

375 questions · 120 min timed mode · Google Cloud · Updated 2026

Vertex AI end to end: frame problems, train models, ship them, monitor the drift.

What you'll be tested on

Sample PMLE questions

You are building an ML model to detect anomalies in real-time sensor data. You will use Pub/Sub to handle incoming requests. You want to store the results for analytics and visualization. How should you configure the pipeline?
  1. 1 = Dataflow, 2 = AI Platform, 3 = BigQuery
  2. 1 = DataProc, 2 = AutoML, 3 = Cloud Bigtable
  3. 1 = BigQuery, 2 = AutoML, 3 = Cloud Functions
  4. 1 = BigQuery, 2 = AI Platform, 3 = Cloud Storage
Show answerA — 1 = Dataflow, 2 = AI Platform, 3 = BigQuery
The correct architecture is Dataflow, AI Platform, BigQuery. Dataflow is the serverless streaming processing service that consumes messages from Pub/Sub in real time, so it fits the real-time sensor ingestion requirement. AI Platform serves the trained anomaly detection model for predictions on the streaming data, and BigQuery is the analytics data warehouse that stores results for analysis and visualization with tools like Looker Studio. Dataproc is for Hadoop/Spark clusters, not managed streaming. Bigtable is a low-latency NoSQL store, not an analytics warehouse, and Cloud Storage is object storage without query analytics, making the other options incorrect.
Your organization wants to make its internal shuttle service route more efficient. The shuttles currently stop at all pick-up points across the city every 30 minutes between 7 am and 10 am. The development team has already built an application on Google Kubernetes Engine that requires users to confirm their presence and shuttle station one day in advance. What approach should you take?
  1. 1. Build a tree-based regression model that predicts how many passengers will be picked up at each shuttle station. 2. Dispatch an appropriately sized shuttle and provide the map with the required stops based on the prediction.
  2. 1. Build a tree-based classification model that predicts whether the shuttle should pick up passengers at each shuttle station. 2. Dispatch an available shuttle and provide the map with the required stops based on the prediction.
  3. 1. Define the optimal route as the shortest route that passes by all shuttle stations with confirmed attendance at the given time under capacity constraints. 2. Dispatch an appropriately sized shuttle and indicate the required stops on the map.
  4. 1. Build a reinforcement learning model with tree-based classification models that predict the presence of passengers at shuttle stops as agents and a reward function around a distance-based metric. 2. Dispatch an appropriately sized shuttle and provide the map with the required stops based on the simulated outcome.
Show answerC — 1. Define the optimal route as the shortest route that passes by all shuttle stations with confirmed attendance at the given time under capacity constraints. 2. Dispatch an appropriately sized shuttle and indicate the required stops on the map.
Because users already confirm their presence and station a day in advance through the GKE application, the pickup demand is known rather than something to predict. The problem becomes an optimization task: find the shortest route visiting only the stations with confirmed riders, subject to vehicle capacity constraints. This is a classic vehicle routing problem, not a machine learning problem, so building regression, classification, or reinforcement learning models adds unnecessary complexity and prediction error when exact confirmed data exists. Option C correctly frames the task as constraint-based route optimization, then dispatches an appropriately sized shuttle with the required stops marked on the map.
You were asked to investigate failures of a production line component based on sensor readings. After receiving the dataset, you discover that less than 1% of the readings are positive examples representing failure incidents. You have tried to train several classification models, but none of them converge. How should you resolve the class imbalance problem?
  1. Use the class distribution to generate 10% positive examples.
  2. Use a convolutional neural network with max pooling and softmax activation.
  3. Downsample the data with upweighting to create a sample with 10% positive examples.
  4. Remove negative examples until the numbers of positive and negative examples are equal.
Show answerC — Downsample the data with upweighting to create a sample with 10% positive examples.
With less than 1% positive examples, the model cannot learn the minority class. The recommended approach is to downsample the majority (negative) class and upweight the downsampled examples so the effective class distribution is preserved, yielding roughly 10% positives as Google recommends for severe imbalance. Downsample and upweight keeps training statistically sound while letting the model converge. Simply generating or removing examples until classes are equal (options A and D) distorts the true distribution and discards useful data or fabricates it improperly. Changing the architecture to a CNN with max pooling (B) addresses feature extraction, not class imbalance, so it does not solve the convergence problem.

Access plans

AccessPrice
3 months$8.99$3.99
1 year$14.99$8.99
Lifetime$24.99$14.99
Practice PMLE now →
Free preview inside — try 5 questions before you pay anything.

FAQ

How many practice questions are in this PMLE bank?
375 questions covering the current PMLE Professional Machine Learning Engineer syllabus, every one with the correct answer and an explanation.
How long is the real PMLE exam?
The official PMLE exam gives you 120 minutes. Our timed exam mode uses the same limit so the pace feels familiar.
What does PMLE access cost?
Plans start at $3.99 for 3 months. One payment, no subscription — and far cheaper than retaking the real exam.