AI-300 mock exam

Question 1, Design and implement an MLOps infrastructureYou must recreate an Azure Machine Learning workspace identically in dev, test and prod. What do you use?

• AA shared wiki page
• BA copy of the dev workspace's notebooks
• CManual steps in the portal
• DBicep templates deployed through a pipeline

Question 2, Implement machine learning model lifecycle and operationsYour team wants to compare parameters and metrics across hundreds of training runs. Which tool does the blueprint expect?

• AAzure Key Vault
• BA spreadsheet updated by hand
• CAzure Service Bus
• DMLflow experiment tracking

Question 3, Implement machine learning model lifecycle and operationsA new model version must reach production with low risk and quick recovery. What do you do?

• AReplace the old deployment in one step during a quiet period
• BAsk a small group of users to report any problems they see
• CRoll out progressively to a share of traffic with a rollback path
• DDeploy on a Friday when traffic is lower and the team can watch it

Question 4, Implement machine learning model lifecycle and operationsModel accuracy is falling because incoming data no longer looks like the training data. What is this, and what should be automated?

• AQuota exhaustion, fixed by requesting a limit increase on the workspace
• BData drift, with monitoring that triggers an alert or retraining
• COverfitting, fixed by lowering the learning rate and retraining once
• DA networking fault, fixed by adding a firewall rule for the data source

Question 5, Design and implement a GenAIOps infrastructureA customer-facing app has steady, high token volume and needs predictable latency. Which deployment option should you evaluate?

• AProvisioned throughput units
• BA free trial deployment
• CA notebook running on a developer machine
• DA manual batch job once a day

Question 6, Design and implement a GenAIOps infrastructureA prompt change caused a production regression, and nobody can tell what changed. What practice prevents this?

• ASharing prompt changes over chat so the whole team sees them
• BKeeping the current prompt in code comments next to each call
• CVersioning prompts in Git with review and evaluation
• DEditing prompts directly in the portal with a change note

Question 7, Implement generative AI quality assurance and observabilityWhich metric tells you whether an answer is supported by the retrieved context?

• AFluency
• BGroundedness
• CCoherence
• DThroughput

Question 8, Optimize generative AI systems and model performanceRetrieval misses documents that contain exact product codes, though semantic matches work well. What do you try?

• AMore replicas of the retrieval endpoint
• BHybrid search that adds keyword matching
• CA larger model for the generation step
• DA higher similarity threshold on vector search