Describe AI MLOps/LLMOps Ecosystems and General Use CERTIFIED-AI-EXPERT Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A fintech company is building an LLM-based system to answer customer queries about their accounts. The system must handle sensitive data and comply with data protection regulations. Which infrastructure consideration is most important for this LLMOps deployment?

   • ARelying on the model provider's default security settings without additional configuration.
   • BUsing the largest available model to maximize accuracy, regardless of data residency requirements.
   • CEnsuring the model is deployed in a secure environment with proper access controls and data encryption.
   • DDeploying the model on shared infrastructure to reduce costs, even if it means data may be processed in other regions.
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A financial services firm is deploying a fraud-detection model that must be retrained monthly with new transaction data. The data science team wants to automate the pipeline from data ingestion through training, validation, and deployment, while also tracking model versions and performance metrics over time. Which approach best aligns with MLOps principles?

   • AHave data scientists manually retrain the model each month and upload the new model file to a shared drive for the operations team to deploy.
   • BUse a notebook environment to retrain the model and manually copy the metrics into a spreadsheet for record keeping.
   • CDeploy the model once and rely on the initial training data, updating only the code that processes incoming transactions.
   • DSet up a CI/CD pipeline that automatically retrains the model on a schedule, runs validation tests, and promotes the model to production with version tracking and monitoring.
   Reveal Solution Discussion Select an answer first
4. 18expert · hard

   A multinational bank is adopting MLOps for its credit-scoring models. The bank must comply with strict audit requirements, including the ability to reproduce any model's predictions at a given point in time. The data science team uses a mix of on-premises and cloud resources. Which approach best ensures reproducibility while meeting audit needs?

   • ARely on the cloud provider's automatic logging and assume it captures all necessary information.
   • BUse a versioned data catalog, containerized training environments, and a model registry that records the exact code, data, and hyperparameters for each run.
   • CKeep all training data in a single database and use a fixed training script that is updated manually.
   • DStore the final model files in a shared network drive and document the training process in a wiki.
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which scenario is a common use case for applying LLMOps practices?

   • ABacking up files to an external hard drive
   • BDeploying a chatbot that answers customer queries using a large language model
   • CConfiguring a firewall to block network traffic
   • DStoring customer records in a relational database
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A telecom company is deploying a churn prediction model. The model is retrained monthly, but the data science team notices that the model's performance degrades shortly after deployment. They suspect the issue is due to changes in customer behavior. Which MLOps practice would most directly address this problem?

   • AIncreasing the frequency of manual model evaluations by the data science team.
   • BImplementing continuous monitoring for data drift and model performance, with automated retraining triggers.
   • CReducing the amount of training data to make the model less dependent on historical patterns.
   • DSwitching to a more complex model that is less sensitive to changes in customer behavior.
   Reveal Solution Discussion Select an answer first