Machine Learning Implementation and Operations Questions
Practice questions for Machine Learning Implementation and Operations topic in AWS Certified Machine Learning - Specialty. 43 questions covering this domain.
An ML team wants to implement an end-to-end MLOps pipeline that automatically triggers data preprocessing, model training, model evaluation, and condi...
A company wants to gradually roll out a new recommendation model (Model B) while keeping the existing model (Model A) in production. They want to rout...
A team has trained a PyTorch computer vision model on SageMaker and wants to deploy it on an embedded ARM processor at the edge with optimized inferen...
A recommendation system uses SageMaker Feature Store for user-profile features. The model requires millisecond-latency feature reads for real-time inf...
A company has trained a SageMaker model and needs to run inference on 10 million records stored in Amazon S3. They do not need a persistent endpoint; ...
A medical imaging company needs to run inference on high-resolution MRI scan payloads up to 1 GB each using a SageMaker endpoint. Processing takes up ...
A startup deploys an ML model that receives unpredictable traffic — no requests for hours, then sudden bursts. They want to minimize cost by not payin...
A company deployed a fraud detection model to a SageMaker real-time endpoint six months ago. They suspect the input feature distribution has shifted o...
A SaaS company has trained 500 separate XGBoost models — one per customer — and needs to deploy all of them for online inference. Deploying 500 indivi...
A SageMaker real-time endpoint is experiencing high latency during peak hours because a single instance is overloaded. The team wants the endpoint to ...
A team needs to monitor a deployed SageMaker real-time endpoint for model quality degradation — specifically detecting when the model's prediction acc...
A team wants to deploy a shadow variant alongside their production model variant on a SageMaker endpoint. The shadow variant receives copies of all pr...
An ML team has trained and validated a new model version and wants to register it in a central catalog that tracks version history, approval status, a...
A company uses SageMaker Pipelines to retrain and deploy models automatically when new data arrives. A new pipeline run completes and the model evalua...
A ML team uses SageMaker Pipelines to automate their entire ML workflow. After training, they want to register the model in the Model Registry only if...
A data science team wants to reduce training costs on SageMaker by using spare EC2 capacity. They are training a deep learning model that takes 6 hour...
A SageMaker endpoint serving a real-time model is experiencing increased invocation latency. The operations team sees that the SageMaker:InvocationsPe...
A company deploys a SageMaker real-time endpoint that serves a fraud detection model. They want to compare a new model version against the current mod...
A team deploys a SageMaker Inference Pipeline with three containers: a scikit-learn preprocessor, an XGBoost predictor, and a postprocessing container...
An operations team wants to track all inputs, outputs, and transformation steps from raw data ingestion through feature engineering, model training, a...
Sign in to see all 43 questions
Create a free account to browse all questions — completely free during our launch phase.