AI0-001 Implementing AI Solutions Practice Question

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Streaming responses from a single monolithic model

  Why it's wrong here

  A single monolithic model cannot serve per-user, per-session updates at browsing latency; the pattern needs an event-driven pipeline that scores incrementally as clickstream events arrive. Streaming from one model suits batch or offline scoring, where throughput matters more than freshness.

• ✗

  Batch processing with nightly updates

  Why it's wrong here

  Nightly batch processing refreshes recommendations only once per day, so it cannot reflect browsing behaviour as users interact with the platform. It is tempting because batch pipelines are cheap and simple, and they are correct for offline model training or daily reporting where latency is irrelevant.

• ✗

  Async processing queue with delayed responses

  Why it's wrong here

  An async queue with delayed responses decouples request from result, so the user waits for a later callback rather than receiving recommendations during the browsing session. It is tempting because queues absorb load spikes, and they are correct for long-running jobs such as order fulfilment, not interactive inference.

• ✓

  AI microservices with a REST API

  Why this is correct

  A REST API lets the application call the recommendation model per user interaction, returning fresh ranked results within the request cycle. This request-response pattern satisfies the real-time update constraint, unlike batch scoring, which would serve stale recommendations between scheduled runs.