Q01 - Question
A media company configures a client application to process contracts, product photos, call recordings, and training videos. The application writes the extracted values into a structured database. Which statement correctly describes the output the application receives from Content Understanding?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Each content type requires a separate service deployment, because a single analyzer API cannot accept audio or video data.
- B. Results are returned only as plain text, so the application must parse structured field values from the text block.
- C. The application receives typed field values, Markdown, and metadata based on the analyzer schema and the content.
- D. The application receives the original file with annotations, and field extraction must be added by a separate model.
C is correct.
Explanation: Analysis outputs expose typed fields, Markdown, and metadata based on the analyzer schema and the content, allowing applications to write type-specific values directly into a structured database.
A is incorrect: You can submit binary document, image, audio, or video data to a single analyzer API.
B is incorrect: Extracted fields are returned as typed values, not just plain unstructured text.
D is incorrect: The application receives the extracted or generated field values from the analyzer, so no separate extraction model is required.
Q02 - Question
A development team is preparing a Python client application that calls Azure Content Understanding. The team wants to use the simplest configuration to let the client connect to the service without integrating Microsoft Entra ID. Which configuration option provides the required values for this approach?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. The Microsoft Foundry resource endpoint and one of the API keys
- B. A storage account connection string and a shared access signature
- C. A Microsoft Entra tenant ID and an application client secret
- D. The analyzer schema file path and the Content Understanding Studio URL
A is correct.
Explanation: The API needs a Microsoft Foundry resource endpoint and key as the direct alternative to project access through the Foundry SDK and Microsoft Entra ID.
B is incorrect: A storage account connection string and a shared access signature do not authenticate calls to the Content Understanding API.
C is incorrect: A tenant ID and an application client secret apply to the Microsoft Entra ID path, which the team wants to avoid in this configuration.
D is incorrect: The schema file path and the Studio URL support analyzer development, but they do not provide the endpoint and key the client needs to connect.
Q03 - Question
A marketing team needs original illustrations and logo concepts for a campaign. No existing artwork matches the described concepts. Which approach in Microsoft Foundry meets this requirement?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Use an image-generation model such as gpt-image-1 to create graphical content from natural-language descriptions.
- B. Use an image classification model to assign category labels to existing campaign artwork.
- C. Use an optical character recognition operation to read text from scanned logo files.
- D. Use an object detection model to return bounding boxes for items in stock photography.
A is correct.
Explanation: Microsoft Foundry can use models such as gpt-image-1 to create original graphical content from a description of a desired image, which supports illustrations, photorealistic images, and logos.
B is incorrect: Classification assigns labels to images you already have, so it produces no new artwork for the campaign.
C is incorrect: Optical character recognition extracts text from existing files and cannot produce new illustrations.
D is incorrect: Object detection locates items in existing images and returns positions, not newly generated graphics.
Q04 - Question
You are building a client application that calls the Images API to generate images and save them as files. Which two statements about the request and response apply? Each correct answer presents part of the solution.
Domain: Implement computer vision solutions (10–15%) Type: Multiple choice
- A. The request supplies values such as the model, prompt, count, and size.
- B. The response returns binary Base64 data that you decode and write to an image file.
- C. The response returns a list of catalog identifiers that you resolve to stored images.
- D. The request must include a training dataset before an image can be generated.
A and B are correct.
Explanation: An Images API request supplies a model, prompt, count, and size, and the binary Base64 result can be decoded and written to an image file by your client application.
C is incorrect: The model generates new image data instead of returning identifiers for images stored in a catalog.
D is incorrect: No training dataset is submitted with the request, because the prompt text drives the generated output.
Q05 - Question
A Python application submits scanned invoices to a Content Understanding analyzer. Developers want to avoid writing custom code to repeatedly check whether the long-running analysis has finished. Which option should the developers use?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Call the REST analyze endpoint directly and read the extracted fields from the initial response body.
- B. Upload each invoice to Content Understanding Studio and download the asynchronous results file.
- C. Use the Python SDK's begin_analyze poller and call .result() to wait for the analysis to complete.
- D. Submit only URL-hosted files, because URL-hosted analysis returns results synchronously.
C is correct.
Explanation: The Python SDK handles authentication, request formatting, and automatic polling for asynchronous operations. The begin_analyze poller handles waiting through .result().
A is incorrect: Analysis is asynchronous, so a REST client must submit the request and then manually poll for the analyzer results rather than reading fields from the initial response body.
B is incorrect: Content Understanding Studio is used for interactive analyzer development, not as a programmatic solution to replace custom polling in a Python application.
D is incorrect: Analysis is asynchronous whether you supply a URL-hosted file or binary data.
Q06 - Question
A finance team wants Content Understanding to return specific values, such as supplier name and total amount, from purchase order documents. The team has already created a Foundry resource. What should the team do next?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Define a schema for the information to extract, based on a sample document and a template.
- B. Submit production purchase orders to the analyzer-results endpoint.
- C. Publish a new version of the analyzer before any fields are defined.
- D. Convert every purchase order to an image file before any configuration.
A is correct.
Explanation: An analyzer is built from a schema that defines the information to extract from a content type, and you define that schema from a sample and a template before you build the analyzer.
B is incorrect: You can only analyze new content after the analyzer is built from a schema.
C is incorrect: Versioning applies to an analyzer that already has a defined schema.
D is incorrect: Content Understanding analyzes documents as well as images, so converting the files is not a required step.
Q07 - Question
A logistics company stores thousands of scanned delivery photos and forms. The team wants the content returned as structured data that downstream analytics and search systems can consume. Which approach matches how Azure Content Understanding produces that output?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Define a schema that describes the data you want, and let the service extract it from the images and documents
- B. Upload the files to a storage container and rely on file metadata to build the structured records
- C. Train a translation model so that the text in each image is converted before analysis
- D. Export each image to plain text and run a keyword search to build the records
A is correct.
Explanation: Azure Content Understanding uses generative AI to analyze unstructured content and return structured data. You define a schema that describes the data you want, and the service extracts it from the images and documents so the output can be used for automation, analytics, and search.
B is incorrect: File metadata describes the file properties, not the unstructured values inside the image or document, so it does not produce the structured fields you need.
C is incorrect: Translating text does not define a schema or return the structured fields that analytics and search require.
D is incorrect: Keyword search over plain text does not produce schema-based structured output with defined fields.
Q08 - Question
A developer wants to verify that a model can answer questions about a local image before writing any application code. The developer uses the Microsoft Foundry portal. What should the developer do?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Deploy a text-only model and paste a text description of the image into the chat playground
- B. Write the full chat application first, and then test image prompts in code
- C. Open the chat playground without deploying a model and upload the image
- D. Deploy a multimodal model, such as gpt-4.1-mini, and use the chat playground to upload the local image with a text prompt
D is correct.
Explanation: Image prompts require a deployed multimodal model that supports text and image input. In the chat playground, you can test the model by uploading a local image and pairing it with a text prompt.
A is incorrect: A text-only model doesn't accept image input. A text description doesn't test how the model responds to the actual image.
B is incorrect: The requirement is to test before writing code. The chat playground lets you test image prompts first.
C is incorrect: You must deploy a multimodal model before you can use it to handle prompts that include images.
Q09 - Question
A team must create Content Understanding analyzers automatically as part of a deployment pipeline. The pipeline uses REST calls to interact with the API. Which approach should the team configure the pipeline to use?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Export a trained analyzer from Content Understanding Studio at run time and import it with a GET request.
- B. Submit a JSON analyzer schema with a PUT request, then poll the URL in the Operation-Location response.
- C. Send a content file to the endpoint first and let the service generate an analyzer from the detected fields.
- D. Create the analyzer with a POST request that returns the finished analyzer definition in the synchronous response.
B is correct.
Explanation: With REST, you submit a JSON analyzer schema using PUT and then poll the Operation-Location value returned in the response, because analyzer creation is asynchronous.
A is incorrect: Exporting a trained analyzer and importing it with a GET request does not programmatically create an analyzer in a pipeline.
C is incorrect: Submitting a content file to the service analyzes the content but does not generate an analyzer schema.
D is incorrect: Analyzer creation is asynchronous and does not complete within a single synchronous POST request response.
Q10 - Question
You are designing a web application that lets users request Sora 2 videos from text prompts. You need to choose an interaction pattern that matches how Sora 2 returns generated video. What should you implement?
Domain: Implement computer vision solutions (10–15%) Type: Single choice
- A. Submit the prompt and settings as a generation job, then retrieve the video when the job completes.
- B. Send the prompt in a synchronous call and hold the user request open until the video bytes return.
- C. Send the prompt to the deployment and read the video from the response stream token by token.
- D. Upload the prompt to a storage container and let the deployment write a video back to the same container.
A is correct.
Explanation: Video generation with Sora 2 is an asynchronous process. You submit the prompt and settings, then retrieve the result after generation finishes.
B is incorrect: A synchronous call that blocks until the video is returned does not match the asynchronous generation process.
C is incorrect: Sora 2 returns generated video content after the job completes rather than as streamed text output.
D is incorrect: You submit the generation request to the model and retrieve the completed video; the request is not made by placing prompt files in a storage container.