Question 70
At fictional Brookstream Commerce Alliance, a Snowflake data and AI engineering team is conducting a restricted enterprise language-model deployment under an approved model lifecycle and release-control process. The separately composed technical incident is: A private inference service runs in a container-backed workload, but the deployment plan exposes a development endpoint broadly and allows outbound calls unrelated to the model use case. Assume only these documented facts and use safe, authorized production-change procedures; exact current Snowflake feature syntax and region or account availability require current review. Which repeatable assurance or control is most relevant to preventing this situation?
Explanation
Writer-proposed answer: B (UNVERIFIED)
Preliminary individual option explanations — NOT expert validated
Correct Answer
Add predeployment endpoint exposure scans, identity/egress reviews, resource quotas and periodic service-configuration drift checks. Container-based model hosting still requires workload identity, network policy, secret hygiene and controlled service lifecycle.
Incorrect Answers
This response concerns Embedding model migration consistency, which is different from the stated failure. A private inference service runs in a container-backed workload, but the deployment plan exposes a development endpoint broadly and allows outbound calls unrelated to the model use case.
Model-selection latency and reliability tradeoffs is not the cause of the reported failure: A private inference service runs in a container-backed workload, but the deployment plan exposes a development endpoint broadly and allows outbound calls unrelated to the model use case.
Separate development and production model rollout alone does not establish the needed evidence: Independent access tests reject anonymous and unauthorized endpoint calls, while service logs show only documented network destinations and safe secret handling.