1. You are building a machine learning application that needs to run inference locally on edge devices. You have decided to use AWS IoT Greengrass to deploy and manage the machine learning models on edge devices. You want to ensure that the edge devices have the necessary resources to perform inference efficiently, while also limiting the memory and CPU utilization. What is the best way to achieve this?

A) Use AWS IoT Analytics to deploy machine learning models on edge devices and configure the resource limits for each Lambda function. This approach is not suitable for running inference on edge devices as it requires data to be sent to the cloud for processing, which can result in latency and privacy issues.
B) Use AWS IoT Greengrass Machine Learning with AWS IoT Greengrass Core to deploy machine learning models on edge devices and configure the resource limits for each Lambda function. You can use the AWS IoT Greengrass runtime environment to optimize the performance of the machine learning models.
C) Use AWS IoT Greengrass to deploy machine learning models on edge devices without configuring the resource limits. This approach may result in high memory and CPU utilization on edge devices, which can impact the performance of the machine learning models.
D) Use AWS IoT Greengrass Stream Manager to manage the data streaming from edge devices to the cloud, but use AWS Lambda to deploy machine learning models on edge devices without configuring the resource limits. This approach may result in high memory and CPU utilization on edge devices, which can impact the performance of the machine learning models.



2. You have an Amazon ECS cluster that is running multiple services. You want to ensure that each service can scale independently based on its own metrics. What should you use to achieve this?

A) Service Auto Scaling with target tracking scaling policies.
B) ECS task placement strategies.
C) AWS Fargate capacity providers.
D) Amazon ECS cluster auto scaling.



3. Which of the following statements accurately describe the capabilities of Amazon S3 for machine learning workflows?

A) Amazon S3 enables direct training of machine learning models using built-in algorithms and frameworks such as TensorFlow, PyTorch, and MXNet.
B) Amazon S3 can be used to stream real-time data for machine learning inference using AWS Kinesis or Apache Kafka.
C) Amazon S3 can be used as a data lake to store and process large volumes of structured data for machine learning workloads.
D) Amazon S3 provides integrated data versioning and rollbacks to support the reproducibility of machine learning experiments.



4. In Amazon Machine Learning, what is the difference between a batch prediction and a real-time prediction, and in which scenarios should you use each?

A) Batch prediction is used when processing a large dataset, while real-time prediction is used when processing small datasets.
B) Batch prediction is used for offline processing, while real-time prediction is used for online processing.
C) Batch prediction is used when the data is streaming in real-time, while real-time prediction is used for batch processing.
D) Batch prediction and real-time prediction have the same functionality, and you can use either of them interchangeably.



5. As a data engineer, you are designing a data pipeline to transform and load data from an on-premises database into an Amazon Redshift data warehouse. You want to use AWS Data Pipeline to accomplish this task. Which of the following statements is true regarding the use of AWS Data Pipeline for this scenario?

A) AWS Data Pipeline requires you to write custom code to transform data, so it is not a good fit for this scenario.
B) AWS Data Pipeline provides a graphical interface for defining data transformation and loading tasks, making it easy to set up and manage complex data pipelines.
C) AWS Data Pipeline does not support loading data into Amazon Redshift, so it cannot be used for this scenario.
D) AWS Data Pipeline only supports Amazon S3 as a data source, so it cannot be used for this scenario.
E) AWS Data Pipeline supports on-premises data sources and can be used to move data from on-premises databases into Amazon Redshift.