Prepstellar
Get it on Google Play

AI-103 · Practice set 2 of 13

Foundry Tools at a Glance: 10 practice questions

10 questions · Untimed · Free

10 free AI-103 practice questions on Foundry Tools at a Glance, with an explanation for every answer. Untimed. The full mock exam and the timed version are in the app.

Set 2 · Foundry Tools at a Glance · 10 questions Read the lesson

  1. Question 1 of 10

    Which Foundry Tool is built to translate more than 100 in-use, at-risk, and endangered languages and dialects?

    1. ALanguage
    2. BImmersive Reader
    3. CContent Understanding
    4. DTranslator
    Show the answer

    Choose the service whose job is translation across languages and dialects, including at-risk and endangered ones. Reading support and natural language understanding solve different jobs.

    Next → 1 / 10

  2. Question 2 of 10

    Which Foundry Tool is described as an AI service that detects unwanted content?

    1. AContent Understanding
    2. BCustom Vision
    3. CDocument Intelligence
    4. DContent Safety
    Show the answer

    Choose the service for screening unwanted content in a new app. Media analysis and customized image recognition have different purposes.

    Next → 2 / 10

  3. Question 3 of 10

    An app sends an audio file and needs a transcription and a translation of the speech back at the same time. Which Foundry Tool fits?

    1. ADocument Intelligence
    2. BContent Safety
    3. CSpeech
    4. DLanguage
    Show the answer

    Start with the input and outputs: one audio file must produce written words and translated text. Pick the service that handles both audio tasks together; text analysis and document extraction solve different problems.

    Next → 3 / 10

  4. Question 4 of 10

    A team is starting a new application that must detect unwanted content. Someone suggests Content Moderator because it is still available. What should they use?

    1. AContent Moderator, because it is still available
    2. BContent Safety, because Content Moderator is retiring
    3. CEither one, since both detect unwanted content
    4. DContent Understanding, which replaces both of them
    Show the answer

    Still available is not the same as recommended. Content Moderator is on the retirement list: existing apps can keep it, but new AI applications should not adopt it. Content Safety is the current tool for detecting unwanted content.

    Next → 4 / 10

  5. Question 5 of 10

    What does a Foundry Tools pricing tier specify?

    1. APeak transactions per second, features, and transaction cost
    2. BSupported languages, regions, and how many keys can call it
    3. CTraining data size, custom models, and free container hours
    4. DNumber of projects, model deployments, and the hours of uptime
    Show the answer

    Billing follows transactions: what you pay depends on how many transactions you send with your credentials. So each tier sets three things: the maximum transactions per second, which features are on, and the cost for a predefined number of transactions.

    Next → 5 / 10

  6. Keep the ones you got wrong

    In the app, every question you miss comes back exactly when you’re about to forget it.

    Get it on Google Play

  7. Question 6 of 10

    A workflow receives a mix of media content types and must analyze and comprehend them. Which Foundry Tool is described for that job?

    1. ACustom Vision
    2. BContent Understanding
    3. CImmersive Reader
    4. DDocument Intelligence
    Show the answer

    A mix of media needs a service whose stated scope spans media types. A document specialist or an image specialist covers a narrower kind of input.

    Next → 6 / 10

  8. Question 7 of 10

    Developers need the most flexibility to call Foundry Tools from any language and environment, and they are comfortable writing Python. Which way of working fits?

    1. ACI/CD pipelines in Azure DevOps
    2. BOn-premises containers
    3. CClient libraries and REST APIs
    4. DDesigner-driven tools
    Show the answer

    There is a trade-off between ease and control. Designer-driven tools are quickest to set up but can limit customization, while REST APIs and client libraries give the greatest flexibility from any language and environment, at the cost of more effort and expertise.

    Next → 7 / 10

  9. Question 8 of 10

    A hospital uses a Foundry Tool that supports containers to process sensitive records. Processing must run on its own premises, close to the data, and its prebuilt model already does the job. Which approach fits?

    1. AUse a designer-driven tool to automate cloud calls
    2. BCall its cloud REST API from an on-premises app
    3. CTrain a custom model through GitHub Actions
    4. DRun the supported tool in an on-premises container
    Show the answer

    Location and training are separate choices. The prebuilt model already works, so there is no reason to train one. A supported container brings processing close to the hospital’s data on its premises.

    Next → 8 / 10

  10. Question 9 of 10

    A retailer needs to recognize its own product types in shelf photos. Prebuilt image analysis does not know these products, and the team has a catalog of images with the location of each product marked. Which approach fits?

    1. ASend the photos to Vision and rely on its baseline model
    2. BTrain a custom model on the tagged images
    3. CRun a tool in a container so it learns from local data
    4. DSend the photos to Content Safety to flag the products
    Show the answer

    Out of the box is the default, but here the baseline model has already failed the task. That is when you bring your own data, tagged the way the service expects, and train a custom model. Custom Vision exists to customize image recognition for your business.

    Next → 9 / 10

  11. Question 10 of 10

    A small app team without data science skills must count the people in images from a lobby camera. They need it this week and no customization. Which approach fits?

    1. ADeploy Vision on premises, then train it on the lobby data
    2. BTag lobby images, then train a Custom Vision model
    3. CSend each image to Vision and use its prebuilt analysis
    4. DSet up GitHub Actions to train and release a model
    Show the answer

    Counting people in a frame is exactly the kind of insight Vision returns out of the box. With a tight deadline, no data science skills, and no need to customize, the prebuilt call is the right choice; custom training is for when the baseline falls short.

    Next → 10 / 10

  12. You’ve finished this set

    That’s 10 questions on Foundry Tools at a Glance. In the app the ones you miss come back exactly when you’re about to forget them.

    Back to the course Next lesson: Foundry Agent Service at a Glance Get it on Google Play
Get it on Google Play

The whole course, on your phone

Lessons you can read, audio you can listen to on the way to work, and practice that remembers what you got wrong.

Get it on Google Play

“This app is the ‘One Lake’ for preparing the certification exams. Really worth it.”

Abdelhak Y. Google Play review