AWS Certified Machine Learning – Specialty — free practice exam
80 original practice questions aligned to official exam domains. Run a timed, domain-weighted quiz in your browser — no login required.
Exam domains
| Domain | Weight |
|---|---|
| Domain 1: Data Engineering | 28% |
| Domain 2: Exploratory Data Analysis | 24% |
| Domain 3: Modeling | 36% |
| Domain 4: Machine Learning Implementation and Operations | 12% |
Sample practice questions
- What is the MOST appropriate AWS solution to batch transform for large inference?
- What is the MOST appropriate AWS solution to deploy real-time inference endpoints?
- What is the MOST appropriate AWS solution to detect model drift?
- What is the MOST appropriate AWS solution to distributed training on GPU clusters?
- What is the MOST appropriate AWS solution to ETL for ML datasets?
- What is the MOST appropriate AWS solution to feature store for training and serving?
- What is the MOST appropriate AWS solution to hyperparameter tuning jobs?
- What is the MOST appropriate AWS solution to ingest streaming data for ML?
- What is the MOST appropriate AWS solution to label training data at scale?
- What is the MOST appropriate AWS solution to managed Jupyter for exploration?
- What is the MOST appropriate AWS solution to prepare ML features at scale?
- What is the MOST appropriate AWS solution to profile large tabular datasets?
- What is the MOST appropriate AWS solution to query data lakes for features?
- What is the MOST appropriate AWS solution to store features for reuse?
- What is the MOST appropriate AWS solution to train models with built-in algorithms?
- What is the MOST appropriate AWS solution to visualize dataset bias?
- An audit found gaps: the organization must profile large tabular datasets. Which option addresses this?
- An audit found gaps: the organization must train models with built-in algorithms. Which option addresses this?
- A company needs to batch transform for large inference. Which option meets the requirement with the LEAST…
- A company needs to prepare ML features at scale. Which option meets the requirement with the LEAST implementation…
- For cost and security, the team must distributed training on GPU clusters. What is the recommended approach?
- For cost and security, the team must detect model drift. What is the recommended approach?
- After a design review, leadership requires the team to label training data at scale. Which approach is MOST reliable?
- After a design review, leadership requires the team to ingest streaming data for ML. Which approach is MOST reliable?
Browse all 80 indexed questions for this exam.