AIMentor — AI Expertise Learning Platform

13 questions · pass mark 80% · 20 minute limit. The timer auto-submits at zero.

Time remaining: 20:00

Question 1easy
Why does the enterprise (especially EU) market default to Azure for AI?
Azure models are more accurateData residency, compliance, identity/secrets integration, and managed services — business reasons, not the modelAzure is the only cloud with GPUsIt removes the need for prompts
Question 2medium
In Azure OpenAI, what do you call instead of a raw model id?
A deployment name you created (which pins version, region, quota)The GPU cluster idA vector index nameThe subscription id
Question 3medium
How much of your prompt/agent/RAG code changes when moving to Azure OpenAI?
Everything must be rewrittenAlmost nothing — only the client setup (endpoint, api_version) and calling a deployment nameOnly the embeddings changeYou must switch to a different SDK entirely
Question 4medium
What is Azure AI Search, in course terms?
A model-training serviceA managed vector DB with first-class hybrid (keyword+vector) retrieval and a built-in semantic rerankerA container registryA prompt template store
Question 5hard
With Azure AI Search, which RAG stages do YOU still own? (Select all that apply)
Chunking strategyThe generation promptChoosing the embedding model / setting the tenant filter scopeRunning the hybrid-retrieval and reranking infrastructure
Select all that apply.
Question 6medium
What is Azure AI Foundry, framed against this course?
A single LLMThe umbrella platform exposing your whole course as managed services (model catalog, prompt flow, evals, agent service, content safety, tracing)A billing dashboard onlyA replacement for Python
Question 7hard
A client says 'we're an Azure shop.' What does that mean for your vendor-neutral skills?
You must relearn RAG and agents from scratchThe concepts are unchanged; you locate each one as an Azure service and decide managed-vs-buildYour skills don't apply on AzureYou can only use no-code tools
Question 8medium
How does Azure address the tenant-isolation concern from the Security track?
It ignores itEngine-level metadata filters in AI Search (e.g. filter by tenant_id) scope retrieval at the sourceBy training a model per tenantBy raising temperature
Question 9hard
When is choosing Azure the right call? (Select all that apply)
The client is already on Azure / Microsoft 365They need EU data residency + enterprise complianceThey want managed search/safety/agents and Entra ID/Key Vault integrationThey are already fully on AWS and want the fastest path
Select all that apply.
Question 10medium
Why does content filtering come 'for free' on Azure OpenAI?
It doesn't; you must build itAzure applies built-in, per-deployment content filtering (managed moderation) automaticallyThe model refuses everything by defaultIt only filters embeddings
Question 11hard
What is a pitfall specific to working with the Azure AI platform?
Prompts don't work on AzureChasing shifting product names/API versions (Studio→Foundry) instead of anchoring on the conceptAzure has no vector searchYou can't use open models
Question 12medium
In Azure OpenAI, you call a ____ name rather than a raw model id (one word).
Question 13hard
Which Azure service provides managed hybrid + semantic-reranked retrieval (the managed RAG store)? (two words after 'Azure AI')

0 / 13 answered

Azure AI Platform — Final Exam