# Google ML Engineer (GME) practice exams

> Google ML Engineer practice exams: 241 questions in 5 timed tests, every answer explained.

Page: https://devopscodelab.com/exams/google-ml-engineer/

241 questions in 5 timed practice tests. Each test has about 51 questions and a 120-minute limit. The pass mark is 70%. Results are split across the 6 exam domains.

## Exam at a glance

| Setting | Value |
| --- | --- |
| Practice tests | 5 |
| Questions in total | 241 |
| Questions per test | About 51 |
| Time limit per test | 120 minutes |
| Pass mark | 70% |
| Question types | Single answer and multiple answer (19 multiple-answer questions) |
| Level | Associate |
| Category | AI & Machine Learning |

These are the settings of our practice tests. Check the official Google Cloud exam page for the current format before you book.

## Domains and weights

| Domain | Weight | Questions in our bank |
| --- | --- | --- |
| Architecting low-code AI solutions | 13% | 32 |
| Collaborating within and across teams to manage data and models | 16% | 38 |
| Scaling prototypes into ML models | 21% | 50 |
| Serving and scaling models | 20% | 48 |
| Automating and orchestrating ML pipelines | 17% | 41 |
| Monitoring AI solutions | 13% | 32 |

## Price and access

- GME practice exams: $12.99 for 12 months of access.
- Free mini exam: 15 questions with explanations (free account): https://devopscodelab.com/app/exams/google-ml-engineer/mini/
- All-Access subscription includes every exam: https://devopscodelab.com/pricing/
- 10 free questions with answers: https://devopscodelab.com/exams/google-ml-engineer/free-practice-questions/

## Sample questions

### Question 1 of 3

A company wants an internal assistant that answers employee questions grounded in 40,000 PDF policy documents stored in Cloud Storage. Answers must cite the source document, and the team wants to avoid building its own chunking, embedding, and retrieval pipeline. What should the ML engineer do?

- A. Generate embeddings for each page with a text embedding model, store them in Vector Search on Agent Platform, and write retrieval and prompt logic in Cloud Run
- B. Fine-tune a Gemini model with supervised tuning on question-answer pairs generated from the 40,000 PDFs so the knowledge is stored in the weights
- C. Create an Agent Search data store over the Cloud Storage bucket and use a search app with generative answers that return citations
- D. Load the PDFs into BigQuery, run ML.GENERATE_EMBEDDING on the text, and use VECTOR_SEARCH results as context in a hand-written prompt template

**Answer:** C

**Explanation:** Agent Search (formerly Vertex AI Search), part of Agent Builder, ingests unstructured documents, handles parsing, chunking, indexing, and retrieval, and can return grounded summaries with citations, so no custom RAG pipeline is needed. Vector Search and BigQuery vector search both work but require the team to build chunking, embedding, and prompting themselves. Supervised tuning does not provide reliable citations and goes stale as policies change.

### Question 2 of 3

Several teams compute the same customer features with slightly different SQL, causing inconsistent model results. Features already live in BigQuery, and two models need low-latency online lookups. What should the ML engineer do to consolidate features?

- A. Register the BigQuery feature tables in Agent Platform Feature Store as feature groups and serve them through an online store
- B. Copy the feature tables into Memorystore for Redis nightly and document the column meanings in a team wiki so models can read them with low latency
- C. Export each team's features to Cloud Storage as Parquet files and load them into every training job directly
- D. Publish a shared BigQuery view per feature and have the online models query the views directly at prediction time

**Answer:** A

**Explanation:** Agent Platform Feature Store (in Gemini Enterprise Agent Platform, formerly Vertex AI) uses BigQuery as its offline source, so existing tables can be registered as feature groups once and served with low latency from an online store, giving every team the same definitions. Shared views fix consistency but querying BigQuery at request time is too slow for online lookups. A Redis copy and Parquet exports duplicate data and leave no governed feature registry.

### Question 3 of 3

An ML team uses Experiments on Agent Platform. A data scientist trains a scikit-learn model in a Workbench notebook and wants each run's parameters, metrics, and the resulting model artifact to be comparable in the console. What is the simplest way to do this?

- A. Create an Agent Platform TensorBoard instance and write scikit-learn scores as TensorFlow summary events inside the notebook's training loop
- B. Write metrics into a BigQuery table after every run and build a Looker Studio dashboard over that table
- C. Call aiplatform.init with the experiment name, then start_run, log_params, and log_metrics in the notebook
- D. Save each trained model to Cloud Storage with the hyperparameters encoded in the object name

**Answer:** C

**Explanation:** The Agent Platform SDK's experiment APIs record parameters and metrics for each run, and runs are compared side by side in the Experiments console and backed by Agent Platform ML Metadata (formerly Vertex ML Metadata). A BigQuery dashboard or object naming conventions are manual and give no lineage. TensorBoard is designed for time-series training curves, and forcing scikit-learn scores into summary events adds work without run comparison.


## Study plans

### Two weeks

1. Day 1: take practice test 1 in exam mode to get a baseline.
2. Days 2 to 5: domain mode on Scaling prototypes into ML models and Serving and scaling models, the two most heavily weighted domains.
3. Day 6: practice test 2 in exam mode, then the mistakes quiz.
4. Days 7 to 12: practice tests 3 to 5, one every other day, each followed by its mistakes quiz.
5. Days 13 and 14: final mode. Book the exam when your readiness stays above the pass mark.

### Four weeks

1. Week 1: one practice test in practice mode, reading every explanation. Note your two lowest domains.
2. Week 2: domain mode on those two domains, then a second practice test in exam mode.
3. Week 3: practice tests 3 and 4 in exam mode, with the mistakes and flagged quizzes after each.
4. Week 4: practice test 5 and final mode until your readiness holds above the pass mark.

## Questions about the GME practice exams

**Are these real GME exam questions?**

No. We write every question ourselves and map it to the published GME exam objectives. We do not use leaked or recalled questions.

**Do the time limit and pass mark match the official exam?**

They match the settings of our practice tests. Exam providers change their format from time to time, so check the official exam page before you book.

**Can I try it before I pay?**

Yes. The free mini exam has 15 questions with full explanations and needs no account.

**How long do I keep access?**

A single-exam purchase gives you 12 months of access. All-Access includes this exam for as long as your subscription runs.

**What happens after I finish a test?**

You get a score, a pass or fail against the pass mark, a score per domain and every answer with its explanation.

Last updated November 2026.

DevOpsCode Lab is not affiliated with Google Cloud. GME and related names are trademarks of their owners. We write every question ourselves; these are not real exam questions.
