Skip to content

Deployment and Orchestration of ML Workflows Questions

Practice questions for Deployment and Orchestration of ML Workflows topic in AWS Certified Machine Learning Engineer - Associate. 44 questions covering this domain.

44 questions12 easy20 medium12 hard
Q1
easy

An endpoint receives infrequent and unpredictable traffic with long idle periods, and the application can tolerate cold starts. Which SageMaker hostin...

Q2
medium

A company wants AWS to benchmark a model and recommend the endpoint configuration that gives the best performance at the lowest cost for either real-t...

Q3
easy

A customer-facing application needs interactive, low-latency predictions from a SageMaker-hosted model. Which hosting option should the team choose?

Q4
hard

A serverless endpoint must still respond within milliseconds during predictable traffic periods, and the team wants to minimize cold starts. What shou...

Q5
hard

An inference pipeline is already running in production, and the team needs to change the container sequence in the pipeline. What is the correct updat...

Q6
medium

A SaaS provider needs to host many models that use the same ML framework and are accessed unevenly, with some models invoked infrequently. The applica...

Q7
medium

An ML platform team wants a purpose-built SageMaker orchestration service to automate ML development workflows without managing orchestration infrastr...

Q8
medium

A team wants a single deployed model that chains preprocessing, prediction, and postprocessing across multiple containers. Which SageMaker capability ...

Q9
medium

A data science team needs predictions for an entire dataset and does not want to keep a persistent endpoint running. They might also preprocess the da...

Q10
easy

A team must accept individual inference requests up to 1 GB and some requests can take close to an hour to process. Which SageMaker hosting option sho...

Q11
hard

A team is considering multi-model endpoints for hundreds of models, but several of the models have consistently high transaction rates and large memor...

Q12
hard

A SageMaker endpoint requires very low cold-start time for ML inference of small Python-based models with sub-second latency targets and millions of r...

Q13
medium

A team needs to gradually shift inference traffic to a new SageMaker model variant on the same endpoint to evaluate performance before fully cutting o...

Q14
medium

A team wants to compare a new SageMaker model variant against the production model using live traffic without exposing customers to the new model's re...

Q15
hard

An MLOps team wants to automatically promote a registered model package to production only when CodePipeline approval is granted and pipeline tests su...

Q16
easy

Which AWS service provides a fully managed registry for ML container images including XGBoost, scikit-learn, TensorFlow, and PyTorch SageMaker images?

Q17
hard

A SageMaker model needs deployment to remote edge devices that run inference offline and synchronize prediction logs back when online. Which AWS servi...

Q18
easy

Which AWS Step Functions feature is recommended for orchestrating SageMaker training, processing, and batch transform jobs as part of an ML workflow?

Q19
medium

An ML team wants Amazon EventBridge to trigger a SageMaker Pipelines execution when a new training dataset arrives in Amazon S3. Which combination imp...

Q20
medium

An ML platform team wants infrastructure-as-code definitions for SageMaker resources and to track changes across environments. Which AWS combination i...

Sign in to see all 44 questions

Create a free account to browse all questions — completely free during our launch phase.