AI-900 Practice Question: Describe features of generative AI workloads on Azure

Answer analysis

Option-by-option breakdown

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

• ✓

  Create Azure Face resource and get endpoint/key, then load or reference the image, then send detection request to Face API, then parse the detected face data.

  Why this is correct

  Provisioning the Azure Face resource first is mandatory because the REST API call requires a valid endpoint URL and subscription key for authentication; after those credentials exist, you must supply the image either as binary data or a public URL so the service can access it. Sending the detection request then invokes the HTTP POST to the /detect endpoint, and parsing the response is last because the JSON payload with face rectangles, landmarks, and attributes is only available after the service returns a 200 status. This sequence mirrors the actual dependency chain: credentials → input → operation → output.

• ✗

  Load or reference the image, then create Azure Face resource and get endpoint/key, then send detection request to Face API, then parse the detected face data.

  Why it's wrong here

  This order incorrectly assumes you can prepare the image independently of the service, but without an Azure Face resource you will have no endpoint or subscription key to authenticate the eventual API call, so the image preparation step is useless in the absence of credentials. The service must be provisioned and its keys retrieved before any code can reference the endpoint, even if you already have a local path or URL ready. Placing image loading first does not break the overall workflow as severely as other errors, but it still violates the prerequisite that resource creation occurs before any interaction with the Face API.

• ✗

  Create Azure Face resource and get endpoint/key, then parse the detected face data, then load or reference the image, then send detection request to Face API.

  Why it's wrong here

  This sequence is illogical because you cannot parse detected face data before a detection request has been made and a response returned; 'parsing' in this context means interpreting the JSON returned by the Face API, which does not exist yet. Additionally, the image must be loaded or referenced before the API call, since the service needs the image bytes or URL as part of the request payload. Even though resource creation is correctly first, inserting parsing before both the image and the request means the order has a fatal logical gap.

• ✗

  Send detection request to Face API, then create Azure Face resource and get endpoint/key, then load or reference the image, then parse the detected face data.

  Why it's wrong here

  This option fails because sending a detection request without first creating the Azure Face resource means there is no endpoint or subscription key to include in the HTTP headers, so the request will be rejected with a 401 or 404 before any image is even processed. The image also needs to be loaded or referenced before the call, because the /detect endpoint requires a body containing the image data, or a URL in a JSON parameter. While parsing is correctly placed after the request, the first three steps are impossible to execute in that order.

Go deeper

Related to this question