Designing Foundry Infrastructure: 10 practice questions

Question 1 of 10

When an organization plans its Foundry topology for production workloads, what should be the default stance?

1. AColocation by default, with isolation as the exception
2. BIsolation by default, with colocation as an exception
3. COne shared project for every production workload
4. DIsolation only for workloads that fine-tune models

Question 2 of 10

In a standard agent setup, which bring-your-own resource holds the vector stores that agents create?

1. AAzure Storage account
2. BAzure AI Search
3. CAzure Key Vault resource
4. DAzure Cosmos DB for NoSQL

Question 3 of 10

A user holds the Contributor role on a Foundry resource and can manage it, yet cannot chat with an agent in the portal. What is missing?

1. AA data plane role for development work
2. BA managed virtual network for the project
3. CThe Owner role on the same Foundry resource
4. DA customer-managed key on the Foundry resource

Question 4 of 10

Three teams share one Foundry resource with a project each. Every team now needs its own independent fine-tuning. What does the planning guidance recommend?

1. AKeep one project and share the fine-tuned models
2. BRun fine-tuning in each team's own project
3. CMove each team onto a separate Foundry resource
4. DGrant each team project-level RBAC on training

Question 5 of 10

You are sizing the Azure Cosmos DB for NoSQL account for a standard agent setup. Which requirement must it meet?

1. AAt least 3000 RU/s total, provisioned or serverless
2. BAt least 3000 RU/s total, provisioned mode only
3. CAt least 1000 RU/s total, provisioned mode only
4. DAt least 1000 RU/s total, serverless mode only

Question 6 of 10

An existing project in a standard agent setup must start storing agent state in a different Cosmos DB account. How do you make the change?

1. AEdit the managed Cosmos DB connection in place
2. BOverride the account defaults from the project
3. CDelete and recreate the project with new settings
4. DUpdate the capability settings on the project

Question 7 of 10

In a shared Foundry resource, every team receives roles at project scope only. A new use case needs speech-to-text. What should the platform team plan for?

1. AA project-scope role that covers speech calls
2. BAn account-level connection to a Speech service
3. CA new project dedicated to the speech use case
4. DRoles at resource scope, or a separate resource

Question 8 of 10

An innovation team wants many small prototypes on non-production data. Each prototype needs its own workspace, but teams should reuse model deployments and pre-connected tools without repeating infrastructure setup. Which setup fits?

1. AOne resource per team, each with one project
2. BOne Foundry resource, one project per use case
3. COne Foundry resource for each prototype idea
4. DOne project that all prototype teams share

Question 9 of 10

A company's prototypes already run in one Foundry account on the basic agent setup. A new regulated workload must keep all its agent data in the company's own Azure resources, and the prototypes must stay untouched. What is the recommended layout?

1. ACapability settings applied to the existing prototype projects
2. BA separate Foundry account and project on standard setup
3. CThe basic setup kept, since it stores data in your own tenant
4. DA standard-setup project inside the existing Foundry account

Question 10 of 10

A Bicep deployment with agent capability settings succeeds. The deploying principal held Storage Blob Data Contributor and Cosmos DB Operator. Now agents fail with 403 Forbidden whenever they read or write files. What fixes it?

1. AGrant the deploying principal Storage Account Contributor
2. BGrant the project managed identity its blob container roles
3. CRaise the Cosmos DB account throughput above 3000 RU/s
4. DAssign the Foundry User role to each developer on the project