Determine the Differences Between Jupyter Notebooks and Pipelines CERTIFIED-AI-EXPERT Practice Questions (Page 3)

Questions 11–15

1. 11expert · hard

   A company has a model that was developed in a Jupyter notebook. The model must now be integrated into a real-time API that serves thousands of requests per second. The API must be highly available and scale automatically based on traffic. The team is considering whether to deploy the notebook directly or refactor it. What is the most important consideration?

   • AThe notebook can be deployed directly if it is wrapped in a REST API framework like Flask.
   • BThe notebook can be deployed directly if it is converted to a Python script and run on a powerful server.
   • CThe notebook can be deployed directly using a tool like Voilà, which turns it into a web application.
   • DThe notebook's stateful kernel is not designed for concurrent request handling, so the logic must be refactored into a stateless service.
   Reveal Solution Discussion Select an answer first
2. 12foundation · easy

   What is a key characteristic of a pipeline's execution model?

   • AIt requires a human to trigger each step individually.
   • BIt allows manual intervention between each step to adjust parameters.
   • CIt maintains an interactive session between steps for debugging.
   • DIt runs steps sequentially in a defined order, often automatically.
   Reveal Solution Discussion Select an answer first
4. 13expert · hard

   A team is auditing a machine learning model that was developed in a Jupyter notebook. The auditor asks the team to prove that a specific model version was trained with a specific dataset and hyperparameters. The team is having difficulty providing this evidence. What is the most likely reason for this difficulty?

   • AThe notebook file is too large to be audited, so the team cannot provide it as evidence.
   • BThe notebook's stateful execution model means the final model depends on the order of cell execution, which is not recorded in the notebook file.
   • CThe notebook was run on a different machine, so the team cannot access the training logs.
   • DThe notebook file does not store the code that was used to train the model.
   Reveal Solution Discussion Select an answer first
5. 14expert · hard

   A data scientist has developed a model in a Jupyter notebook. The scientist runs the cells in a non-linear order during development. The final model is saved to a file. The team wants to reproduce the exact model for a production deployment. What is the most significant challenge?

   • AThe model file is not saved in a format that can be loaded by the production environment.
   • BThe notebook was run on a different machine, so the production team cannot access the training data.
   • CThe notebook file is too large to be shared with the production team.
   • DThe notebook does not record the execution order of the cells, so the team cannot know the exact sequence of operations that produced the model.
   Reveal Solution Discussion Select an answer first
6. 15application · medium

   A team of five data scientists is collaborating on a feature engineering project. They are using a shared Jupyter notebook stored in a Git repository. The team is experiencing frequent merge conflicts and difficulty understanding which version of the code produced a particular result. What is the most effective way to address these collaboration issues?

   • AStore the notebook in a shared network drive instead of Git to avoid merge conflicts.
   • BRefactor the shared logic into modular Python scripts and use the notebook only for lightweight visualization and exploration.
   • CUse a notebook diffing tool that ignores cell outputs and only tracks the source code.
   • DHave only one person edit the notebook at a time and merge changes manually.
   Reveal Solution Discussion Select an answer first