Search...
Master key concepts with interactive flashcards
What is the difference between SageMaker Feature Store Online Store and Offline Store?
Click to reveal answer
Online Store: DynamoDB-backed, < 10ms real-time retrieval via GetRecord API — used for real-time inference. Offline Store: S3-backed, queryable via Athena/Spark — used for batch training data generation. Both can be enabled simultaneously; ingested features are written to both stores.
Click to show question
AWS Machine Learning Engineer - Associate is a AWS certification. These facts come from the current exam guide and are the numbers to plan around.
| Domain | Weight | What to focus on |
|---|---|---|
| Data Preparation for Machine Learning (ML) | 28% | AWS Glue for ETL and data transformation, Amazon S3 data lakes and dataset management, SageMaker Data Wrangler for data preparation |
| ML Model Development | 26% | SageMaker built-in algorithms (XGBoost, Linear Learner, etc.), SageMaker Training Jobs and hyperparameter tuning, Distributed training strategies (data parallelism, model parallelism) |
| ML Solution Monitoring, Maintenance, and Security | 24% | SageMaker Model Monitor for data quality and drift detection, CloudWatch metrics and alarms for ML infrastructure, SageMaker Clarify for bias detection and explainability |
| Deployment and Orchestration of ML Workflows |
| 22% |
| SageMaker Endpoints for real-time inference, SageMaker Batch Transform for batch inference, SageMaker Serverless Inference for variable traffic |
Yes. You can start MLA-C01 practice and flashcards free without creating an account. A free account keeps your progress, and the full bank (300+ questions) unlocks with a subscription.
They cover all 4 domains of the MLA-C01 exam: Data Preparation for Machine Learning (ML) (28%), ML Model Development (26%), ML Solution Monitoring, Maintenance, and Security (24%), Deployment and Orchestration of ML Workflows (22%). Each section is weighted in the same proportion as the real exam, so nothing high-value is buried.
The MLA-C01 exam has approximately 65 questions in 130 minutes, with a passing score of 720. Use the practice questions to confirm you are consistently above that line before booking the real exam.
Study in short, spaced sessions: a card you have not seen for two days is worth more than a card you reread immediately. Filter by Data Preparation for Machine Learning (ML) first, then mix domains so you practise identifying the topic before recalling the definition.
They are the recall half of preparation. Pair the cards with practice questions and at least one timed practice exam — candidates who only review cards usually recognise terms but miss scenario questions.