MLA-C01 Deployment and Orchestration of ML Workflows Practice Test 6 — 15 Questions

Question 1

A company wants to deploy a PyTorch model on SageMaker using the NVIDIA Triton Inference Server for GPU acceleration. They have an existing Triton configuration. Which approach should they take?

Accepted Answer

Use the SageMaker Triton Inference Server container from the Deep Learning Containers. AWS provides a pre-built SageMaker Triton Inference Server container as part of the Deep Learning Containers (DLCs), which is optimized for GPU acceleration and supports the existing Triton configuration without modification. This container integrates directly with SageMaker hosting endpoints, enabling seamless deployment of PyTorch models with Triton's features like dynamic batching and model concurrency.

Answer

Use SageMaker Neo to compile the model for Triton

Answer

Package Triton as a custom container and use SageMaker batch transform

Answer

Use the standard SageMaker PyTorch container and install Triton at runtime

Question 2

A team needs to deploy a model that has compliance requirements to log all inference requests and responses for auditing. The model will be served using a real-time endpoint. How can they achieve this without custom code?

Accepted Answer

Enable SageMaker Data Capture on the endpoint. SageMaker Data Capture is the native, no-code feature that automatically logs inference requests and responses for real-time endpoints. It captures payload data to an S3 bucket without requiring any custom code, directly meeting the compliance requirement for audit logging.

Answer

Add a custom Lambda function using a container

Answer

Use SageMaker Debugger to monitor inference

Answer

Enable CloudTrail for the endpoint

Question 3

A company wants to version and track ML models, with an approval workflow for promoting models from staging to production. Which SageMaker feature should they use?

Accepted Answer

SageMaker Model Registry. SageMaker Model Registry is the correct choice because it provides a centralized repository to catalog, version, and manage ML models, and it supports approval workflows (e.g., PendingApproval, Approved, Rejected) to promote models from staging to production. This directly addresses the requirement for version tracking and an approval gate for model promotion.

Answer

SageMaker Model Monitor

Answer

SageMaker Experiments

Answer

SageMaker Pipelines

Question 4

A startup wants to deploy a model that has variable traffic patterns, with some periods of no traffic and occasional spikes. They want to pay only for what they use and do not want to manage instances. Which SageMaker inference option should they choose?

Accepted Answer

Serverless inference. Serverless inference is the correct choice because it automatically scales to zero during periods of no traffic and scales up to handle spikes, charging only for the compute time used. This eliminates the need to manage underlying instances, making it ideal for variable and intermittent traffic patterns.

Answer

Batch transform

Answer

Real-time endpoint with auto-scaling

Answer

Multi-model endpoint

Question 5

A company wants to deploy a new model using a canary deployment strategy on SageMaker. Which two actions should they take? (Select TWO.)

Accepted Answer

Create a new endpoint with two production variants. To implement canary deployment, create two production variants (current and new) with initial traffic weights (e.g., 95% and 5%), then update the endpoint to gradually shift traffic. Using endpoint update with routing config adjusts traffic weights over time.

Answer

Register both models in the Model Registry with 'Approved' status

Answer

Use SageMaker Model Monitor to compare model performance

Answer

Enable data capture on the endpoint

Question 6

A machine learning engineer is designing a SageMaker Pipeline that includes a training step, a processing step for evaluation, and a condition step to decide whether to register the model. The pipeline should support caching to avoid redundant runs when inputs haven't changed. Which three steps must have caching enabled? (Select THREE.)

Accepted Answer

Training step. For caching to avoid redundant runs, the steps that produce outputs that can be reused must have caching enabled. The processing step (evaluation) and training step both generate outputs that can be cached if their inputs (code, data, hyperparameters) remain the same. The condition step does not produce outputs to cache; it just branches. The RegisterModel step typically registers metadata, but its inputs (model artifact, metrics) may be generated by previous steps; enabling caching on the RegisterModel step can also avoid re-running if the same model artifact is already registered.

Answer

Transform step (if used)

Answer

Condition step

Question 7

A team wants to deploy a single SageMaker real-time endpoint that serves both a PyTorch model for NLP and a TensorFlow model for image classification. Each model requires a different inference container. Which two features can they use together to achieve this? (Select TWO.)

Accepted Answer

Multi-container endpoint. A multi-container endpoint allows running multiple containers (e.g., PyTorch and TensorFlow) on the same endpoint. With inference components, each container can be associated with a specific model, and the routing logic directs requests to the appropriate container based on the model name.

Answer

Multi-model endpoint

Answer

Production variants

Answer

SageMaker Neo compilation

Question 8

A team uses SageMaker Pipelines to retrain a model nightly. They want to skip the training step if the new data is unchanged (same checksum as previous run) to save cost and time. Which pipeline configuration achieves this?

Accepted Answer

Use a ConditionStep that compares the current data checksum to the previous run's checksum, and branch to a NoOp step if unchanged. SageMaker Pipelines' ConditionStep allows you to evaluate a condition—such as comparing the current data checksum to a stored previous checksum—and branch accordingly. If the checksums match, you can route to a NoOp step (which does nothing) instead of executing the training step, thereby skipping the training and saving cost and time. This is the native, recommended pattern for conditional execution in SageMaker Pipelines.

Answer

Enable pipeline caching on the training step

Answer

Use a Lambda step to check data before running the training step

Answer

Set the training step's CacheConfig with a TTL of 24 hours

Question 9

A company needs to serve real-time predictions from a large ensemble of three deep learning models, each requiring different inference environments (PyTorch, TensorFlow, MXNet). Which SageMaker endpoint type supports running multiple inference containers together?

Accepted Answer

Multi-container endpoint. Amazon SageMaker multi-container endpoints allow you to run multiple inference containers (e.g., PyTorch, TensorFlow, MXNet) within a single endpoint, each handling different models or inference environments. This is achieved by deploying multiple containers behind a single endpoint with a serial or direct invocation pattern, enabling real-time predictions from the ensemble without managing separate endpoints.

Answer

Multi-model endpoint

Answer

Real-time endpoint with a single container

Answer

Asynchronous endpoint

Question 10

A data scientist wants to train a model on SageMaker using a custom PyTorch script, then register the best model in the SageMaker Model Registry. The training job is part of a SageMaker Pipeline. Which pipeline step should be used to register the model?

Accepted Answer

RegisterModelStep. The `RegisterModelStep` is specifically designed to create a model resource and register it in the SageMaker Model Registry as part of a pipeline. It takes the training output (e.g., model artifacts from a `TrainingStep`) and packages it with the specified inference image and metadata, then creates a model package group version. This is the correct step for registering a model after training, as it directly integrates with the Model Registry for versioning and approval workflows.

Answer

CreateModelStep

Answer

TrainingStep

Answer

TransformStep

Question 11

A company uses SageMaker Neo to compile a trained model for deployment on edge devices. What is the primary benefit of using Neo?

Accepted Answer

It reduces model size and improves inference speed on target hardware. SageMaker Neo optimizes models for specific hardware architectures (e.g., ARM, Intel, NVIDIA) to achieve faster inference and lower memory footprint.

Answer

It monitors model drift in production

Answer

It automatically retrains the model on new data

Answer

It provides a serverless inference endpoint

Question 12

A team wants to orchestrate a multi-step ML workflow that includes data preprocessing, hyperparameter tuning, model training, evaluation, and conditional deployment to staging or production based on evaluation metrics. The workflow should run on a schedule and track lineage. Which service should they use?

Accepted Answer

SageMaker Pipelines. SageMaker Pipelines is the correct choice because it is purpose-built for orchestrating multi-step ML workflows, including data preprocessing, hyperparameter tuning, model training, evaluation, and conditional deployment. It natively supports scheduling via EventBridge or a cron expression, tracks lineage automatically through SageMaker Experiments and artifact tracking, and allows conditional branching (e.g., deploy to staging or production based on evaluation metrics) using `ConditionStep`.

Answer

Amazon MWAA (Managed Workflows for Apache Airflow)

Answer

AWS Glue workflows

Answer

AWS Step Functions with Lambda functions for each step

Question 13

A team deploys a model on a SageMaker real-time endpoint using an ml.m5.xlarge instance. The model has high latency due to a large neural network. The team wants to reduce latency without changing the model code. Which option should they use?

Accepted Answer

Attach Amazon Elastic Inference to the endpoint. Amazon Elastic Inference attaches a fixed amount of GPU acceleration to an EC2 instance, providing cost-effective acceleration for deep learning inference without needing a full GPU instance.

Answer

Increase the instance size to ml.m5.4xlarge

Answer

Use SageMaker Neo to compile the model

Answer

Switch to a GPU instance like ml.g4dn.xlarge

Question 14

A company uses SageMaker Model Registry to manage model versions. They want to enforce that only models with an 'Approved' status can be deployed to production endpoints. How can they enforce this?

Accepted Answer

Set IAM policies with a condition on sagemaker:ModelVersionStatus. SageMaker Model Registry supports approval workflows. By using IAM policies that conditionally allow deployment only when the model version status is 'Approved', the company can enforce governance.

Answer

Use AWS Lambda to check the model status during deployment

Answer

Use SageMaker Pipelines to deploy only approved models

Answer

Configure SCPs to block deployment of unapproved models

Question 15

A team uses MLflow on SageMaker for experiment tracking. They want to automate the retraining of a model when new training data arrives in an S3 bucket. Which combination of services should they use?

Accepted Answer

EventBridge -> Lambda -> SageMaker Training Job. EventBridge can detect S3 PutObject events and trigger a Lambda function that starts a SageMaker training job, possibly using MLflow for tracking.

Answer

SageMaker Pipelines scheduled trigger every hour

Answer

S3 Event Notifications -> SQS -> SageMaker Training Job

Answer

AWS Step Functions with S3 poller