Developing End-To-End ML Pipelines PROFESSIONAL-MACHINE-LEARNING-ENGINEER Practice Questions (Page 1)

Questions 1–5

1. 1foundation · easy

   What is a key characteristic of using unmanaged services for pipeline orchestration?

   • AThe cloud provider automatically scales and manages all resources
   • BYou are responsible for managing the underlying infrastructure and software
   • CYou can only use open-source tools
   • DThey are always cheaper than managed services
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   A large financial institution has an existing on-premises Ray cluster used for distributed model training. They are migrating to Google Cloud and want to continue using Ray for training while also orchestrating the full ML pipeline (data ingestion, validation, training, evaluation, deployment) in a way that minimizes migration effort. The team has strong Kubernetes expertise and wants to keep control over the Ray cluster configuration. Which approach best meets these requirements?

   • ADeploy Ray on a GKE cluster and use a custom orchestrator (e.g., Argo Workflows) to manage the pipeline steps, calling Ray for distributed training.
   • BUse Cloud Composer to orchestrate the pipeline and run Ray jobs on a Dataproc cluster.
   • CUse Vertex AI Pipelines with a custom container that runs Ray inside a single pod.
   • DMigrate all training to Vertex AI Training and use Vertex AI Pipelines for orchestration.
   Reveal Solution Discussion Select an answer first
4. 3foundation · easy

   Which of the following is an important consideration when designing a custom ML pipeline solution?

   • AScalability, maintainability, and integration with existing systems
   • BAvoiding any use of cloud services to reduce costs
   • CEnsuring the pipeline runs only on a single machine for simplicity
   • DUsing only proprietary technologies to avoid open-source licensing
   Reveal Solution Discussion Select an answer first
5. 4expert · hard

   A research lab runs a complex ML pipeline that involves custom distributed training with a proprietary algorithm. They need to orchestrate the pipeline steps (data prep, training, evaluation) and have a dedicated team that can manage infrastructure. They want to use an open-source orchestration tool that is widely adopted and can run on their existing Kubernetes cluster. Which tool should they use?

   • ACloud Composer
   • BApache Beam
   • CKubeflow Pipelines
   • DVertex AI Pipelines
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   Which managed service is designed for orchestrating ML pipelines on Google Cloud, providing a fully managed environment for building and running pipeline components?

   • ACloud Composer
   • BCloud Storage
   • CVertex AI Pipelines
   • DCloud Functions
   Reveal Solution Discussion Select an answer first