AI-102 Intermediate Quiz - Free Practice Quiz

Question 1 of 60

An Azure AI Engineer is building a Retrieval-Augmented Generation (RAG) solution. Which combination of Azure services MOST effectively enables grounded, factual responses from an Azure OpenAI model?

A Azure Cosmos DB + Azure Translator + Azure OpenAI GPT-4 B Azure Blob Storage + Azure Functions + Azure Bot Service C Azure Custom Vision + Azure AI Speech + Azure OpenAI D Azure AI Search (for document retrieval) + Azure OpenAI (for generation) + Azure AI Document Intelligence (for document ingestion)

1 / 60

Question 2 of 60

An Azure AI Engineer is designing a solution to extract invoice data at scale using Azure AI Document Intelligence. Which model type MOST reduces development effort for standard invoice formats?

A The General Document model with manual field mapping B The Read model for OCR only, then custom NLP for field extraction C A Custom model trained from scratch on 5 sample invoices D The prebuilt Invoice model â€” directly extracting vendor, amounts, line items, and dates without custom training

2 / 60

Question 3 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Language feature extracts named entities and classifies them into predefined types (person, organization, location) from text?

A Key phrase extraction B Custom Text Classification C Language detection D Pre-built Named Entity Recognition (NER)

3 / 60

Question 4 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Language feature is MOST appropriate for classifying customer support tickets into multiple non-exclusive categories simultaneously?

A Single-label custom text classification B Pre-built sentiment analysis C Language detection D Multi-label custom text classification

4 / 60

Question 5 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure OpenAI parameter controls the diversity of responses by restricting sampling to the top cumulative probability mass?

A Temperature B Max tokens C Presence penalty D Top-p (nucleus sampling)

5 / 60

Question 6 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure OpenAI technique MOST effectively improves model responses for a specific task using a small number of labeled examples provided directly in the prompt?

A Adjusting the frequency penalty parameter B Zero-shot prompting with only a system message C Fine-tuning the model on 1,000+ examples D Few-shot prompting â€” including 2â€“5 worked examples in the prompt to guide the model's output format and style

6 / 60

Question 7 of 60

The Azure AI Engineer Associate AI-102 exam covers which Azure AI custom vision feature allows training a model that identifies whether one object is present versus another (binary classification)?

A Object detection model B Multi-label classification model C Semantic segmentation model D Binary image classification model (Compact domain)

7 / 60

Question 8 of 60

The Azure AI Engineer Associate AI-102 exam describes which Azure AI approach as generating text embeddings that can be stored as vectors for similarity search?

A Azure AI Language key phrase extraction B Azure Custom Vision feature embeddings C Azure AI Speech speaker embeddings D Azure OpenAI text-embedding models (e.g., text-embedding-3-large)

8 / 60

Question 9 of 60

On the Azure AI Engineer Associate AI-102 exam, which approach MOST efficiently processes thousands of documents to extract structured data using Azure AI Document Intelligence?

A Processing each document synchronously with the Read API one at a time B Using Azure Custom Vision to classify document types before extraction C Calling the Azure AI Document Intelligence REST API directly from a mobile app D Using Azure AI Document Intelligence in batch mode via the Analyze Document API with async polling, queuing documents through Azure Service Bus

9 / 60

Question 10 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Search feature improves retrieval quality by using an ML model to re-rank initial search results based on semantic relevance?

A Synonym map expansion B Standard BM25 score sorting C Faceted filtering by metadata D Semantic ranker (L2 reranking using a cross-encoder model)

10 / 60

Question 11 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Speech feature enables converting text in one language to spoken audio in another language?

A Speech-to-text B Custom Neural Voice C Speaker recognition D Speech translation (speech-to-speech across languages)

11 / 60

Question 12 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Adaptive Card feature in Bot Framework enables a bot to collect structured user input (forms) within a chat interface?

A Hero Card with button actions B Audio Card for voice responses C Thumbnail Card with image attachment D Adaptive Card with Input elements (text, choice, date inputs) and Submit action

12 / 60

Question 13 of 60

The Azure AI Engineer Associate AI-102 exam covers which technique for reducing hallucinations in Azure OpenAI responses for enterprise knowledge base applications?

A Increasing the temperature parameter to generate more varied responses B Fine-tuning the model on a small set of representative Q&A pairs C Using a smaller, faster model for all queries D Implementing RAG with Azure AI Search to ground the model's responses in retrieved, verified source documents

13 / 60

Question 14 of 60

The Azure AI Engineer Associate AI-102 exam covers which approach MOST effectively handles multilingual user input in an Azure AI Language CLU application?

A Only supporting English and rejecting other language inputs B Building separate CLU projects for each language C Using Azure Translator to translate all input to English before sending to CLU D Training a single CLU project on multi-language utterances and enabling multilingual inference mode

14 / 60

Question 15 of 60

On the Azure AI Engineer Associate AI-102 exam, which approach BEST protects sensitive user data submitted to an Azure OpenAI application from being included in model training by Microsoft?

A Using only GPT-3.5-Turbo which does not log any data B Enabling streaming mode for all API calls C Configuring customer lockbox for Azure OpenAI D Using an Azure subscription with data privacy opt-out â€” Azure OpenAI does not use customer data for model training by default, confirmed by opting into data protection terms

15 / 60

Question 16 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI service monitors AI application inputs and outputs for potential fairness issues and demographic disparities?

A Azure AI Fairness Evaluator via Azure Machine Learning Responsible AI dashboard B Azure Security Center C Azure Content Moderator D Azure Monitor latency alerts

16 / 60

Question 17 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Search configuration enables searching across both keywords and semantic vectors simultaneously?

A Hybrid search â€” combining keyword (BM25) and vector search in a single query with Reciprocal Rank Fusion (RRF) merging B Semantic search without vector fields C Faceted search with custom analyzers D Full-text search with BM25 only

17 / 60

Question 18 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Vision capability processes live video streams to count people crossing a line or dwell time in a zone?

A Azure AI Vision Spatial Analysis (people counting, zone analysis on live video) B Azure Video Indexer (batch video analysis) C Azure Custom Vision real-time classifier D Azure Computer Vision Image Analysis on static frames only

18 / 60

Question 19 of 60

The Azure AI Engineer Associate AI-102 exam identifies which Bot Framework component as enabling a bot to retrieve conversation state stored between turns?

A State management with storage (BotState, ConversationState, UserState) B Channel adapter C Bot Framework activity handler D Azure Service Bus queue

19 / 60

Question 20 of 60

The Azure AI Engineer Associate AI-102 exam covers which Azure deployment configuration enables a bot to respond to voice calls via telephone using Azure AI Speech?

A Direct Line Speech channel connecting Azure Bot Service with Azure AI Speech B Teams channel with voice calling C Azure Communication Services only D Azure Event Grid voice routing

20 / 60

Question 21 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Bot Framework concept represents a reusable, encapsulated unit of conversational logic that manages its own state?

A Dialog (e.g., WaterfallDialog, ComponentDialog) B Channel adapter C Webhook handler D Middleware component

21 / 60

Question 22 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Language feature is MOST appropriate for extracting medical concepts (diagnoses, medications, dosages) from clinical notes?

A Text Analytics for Health â€” a specialized model recognizing medical entities and relationships B Standard named entity recognition pre-built model C Custom NER trained on general text corpora D Sentiment analysis of clinical text

22 / 60

Question 23 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Search configuration prevents unauthorized users from seeing documents they do not have access to in search results?

A Document-level security trimming using security filters (OData filter expressions based on user identity) B Query-time synonym filtering C Field-level encryption D Index-level RBAC read access

23 / 60

Question 24 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Language feature is used to identify and extract dates, times, and quantities from text?

A Pre-built entity recognition (NER) â€” detecting temporal and numeric entity types B Sentiment analysis C Key phrase extraction D Language detection

24 / 60

Question 25 of 60

The Azure AI Engineer Associate AI-102 exam identifies which Azure AI service as enabling content filtering of Azure OpenAI outputs to detect harmful categories?

A Azure AI Content Safety â€” integrated content filtering for Azure OpenAI B Azure Monitor C Azure Policy D Azure Content Moderator (legacy)

25 / 60

Question 26 of 60

The Azure AI Engineer Associate AI-102 exam identifies which Azure Bot Service channel connection enables a bot to participate in Microsoft Teams meetings and respond to user messages?

A Microsoft Teams channel connection in Azure Bot Service B Azure Web Chat channel only C Azure SMS channel via Twilio D Azure Direct Line channel

26 / 60

Question 27 of 60

On the Azure AI Engineer Associate AI-102 exam, which feature of Azure AI Document Intelligence enables training a model to classify documents into types (e.g., invoice vs. receipt vs. contract)?

A Custom Classification model B Read model with regex parsing C Pre-built document models D General Document model with manual rules

27 / 60

Question 28 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure OpenAI safety feature prevents the model from generating content about weapons or violence by default?

A Built-in Azure OpenAI content filtering (harm categories: hate, violence, sexual, self-harm) B Temperature set to 0 C Frequency penalty set to maximum D Prompt injection detection only

28 / 60

Question 29 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure service indexes external data sources (SharePoint, Blob Storage, databases) and makes them searchable using AI enrichment?

A Azure AI Search indexers with cognitive skills B Azure Logic Apps C Azure Event Grid D Azure Data Factory

29 / 60

Question 30 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Content Safety feature detects text that could be used to jailbreak an Azure OpenAI model by overriding its instructions?

A Prompt Shield (jailbreak attack detection) B Hate/violence content filter only C Custom blocklist matching D Azure Defender for AI threat detection

30 / 60

Question 31 of 60

In AI-102 exam preparation, which Azure AI Language feature is MOST appropriate for automatically routing customer support documents to different queues based on document style (formal, conversational, technical)?

A Sentiment analysis B Named entity recognition C Custom text classification for document style D Key phrase extraction for style markers

31 / 60

Question 32 of 60

The Azure AI Engineer Associate AI-102 exam identifies which Azure AI Search feature as automatically enriching indexed documents with AI capabilities (OCR, entity extraction, translation)?

A Query-time semantic ranker B Vector search embedding generation at query time C Cognitive skills pipeline (AI enrichment during indexing) D Synonym map expansion

32 / 60

Question 33 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Vision 4.0 feature generates detailed descriptions of multiple regions within a single image?

A Object detection with bounding boxes B Smart cropping C Dense Captions â€” generating multiple captions for specific image regions D Image classification

33 / 60

Question 34 of 60

According to the Azure AI Engineer Associate AI-102 exam, which approach MOST efficiently manages cost when deploying Azure OpenAI for a batch document summarization workload that is not time-sensitive?

A Provisioned Throughput Units (PTU) for maximum throughput B Standard synchronous API calls for each document C Azure OpenAI Batch API â€” submitting bulk jobs at reduced cost with 24-hour completion SLA D Azure OpenAI with streaming enabled for each document

34 / 60

Question 35 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure OpenAI configuration MOST reduces the risk of the model repeating the same phrases or tokens frequently in a long response?

A Setting temperature to 0 B Setting top-p to 1.0 C Setting a positive frequency_penalty (e.g., 0.5â€“1.0) D Setting max_tokens to a lower value

35 / 60

Question 36 of 60

According to the Azure AI Engineer Associate AI-102 exam, which metric MOST directly measures an Azure AI Search index's relevance quality for a set of known queries?

A Index storage size B Query latency in milliseconds C Normalized Discounted Cumulative Gain (NDCG) â€” measuring ranking quality against ground-truth relevant documents D Number of documents in the index

36 / 60

Question 37 of 60

The Azure AI Engineer Associate AI-102 exam covers which approach for enabling an Azure OpenAI chatbot to maintain conversation history across multiple user sessions?

A Storing conversation history in browser localStorage only B Including the full conversation in every system message C Persisting conversation turns in Azure Cosmos DB and retrieving them to reconstruct the messages array for each API call D Relying on the model's internal memory across API calls

37 / 60

Question 38 of 60

The Azure AI Engineer Associate AI-102 exam covers which Azure AI custom model feature allows a model trained on one language's labeled data to perform inference in other languages?

A Deploying separate custom models for each target language B Azure Translator pre-processing before every inference call C Cross-lingual transfer in Azure AI Language custom models (multilingual training) D Using the pre-built multilingual sentiment model instead of custom models

38 / 60

Question 39 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Speech capability MOST cost-effectively produces spoken audio from text for a high-volume text-to-speech application?

A Custom Neural Voice synthesis for every request B Long audio synthesis API for batch text-to-speech generation C Standard prebuilt neural voice via the Speech SDK for real-time synthesis D Standard REST API synthesis with the most expensive neural voice

39 / 60

Question 40 of 60

On the Azure AI Engineer Associate AI-102 exam, which approach MOST effectively reduces Azure OpenAI API costs for a high-volume summarization application?

A Using the largest available GPT-4 model for all requests B Sending all requests synchronously with maximum max_tokens C Caching responses for identical or near-identical prompts using a semantic cache layer (e.g., Azure Cache for Redis) D Increasing the temperature to generate shorter responses

40 / 60

Question 41 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Language capability extracts opinions about specific aspects of a product or service from review text?

A Key phrase extraction B Language detection C Opinion mining (aspect-based sentiment analysis) D Standard sentiment analysis (document-level)

41 / 60

Question 42 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Document Intelligence model is MOST appropriate for extracting fields from a completely novel document layout with no prebuilt model available?

A The Read model with manual field parsing B The prebuilt General Document model only C Custom extraction model trained on labeled sample documents specific to the novel layout D The prebuilt Receipt model as a fallback

42 / 60

Question 43 of 60

The Azure AI Engineer Associate AI-102 exam identifies which feature of Azure AI Search as enabling users to filter search results by category, date range, or other field values?

A Semantic search B Vector search C Faceted navigation (facets) D Synonym maps

43 / 60

Question 44 of 60

The Azure AI Engineer Associate AI-102 exam covers which Azure AI Vision feature analyzes an image and returns a confidence score for whether it contains adult, racy, or gory content?

A Object detection API B Custom Vision classification C Content moderation via Azure AI Vision (adult/racy content detection) D Optical Character Recognition

44 / 60

Question 45 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure resource provides dedicated, reserved throughput for Azure OpenAI API calls to ensure consistent latency?

A Azure CDN for API caching B Azure API Management rate limiting C Provisioned Throughput Units (PTU) â€” Azure OpenAI Provisioned deployment D Azure Functions with reserved instances

45 / 60

Question 46 of 60

According to the Azure AI Engineer Associate AI-102 exam, which approach MOST securely provides an Azure App Service application with access to Azure AI Services without storing credentials in code?

A Hard-coding the API key in appsettings.json B Managed identity assigned to the App Service, granted Cognitive Services User role â€” no secrets stored C Storing the API key in a GitHub secret accessible to the deployment pipeline D Sharing API keys via a team password manager

46 / 60

Question 47 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Vision feature identifies brand logos in images?

A Object detection B Brand detection (Azure AI Vision brand recognition) C OCR text recognition D Custom Vision multi-label classification

47 / 60

Question 48 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Bot Framework middleware capability intercepts every message before it reaches the bot handler?

A Dialog stack management B Custom middleware registered in the adapter's pipeline C Activity handler base class D Channel adapter configuration

48 / 60

Question 49 of 60

The Azure AI Engineer Associate AI-102 exam identifies which technique as using Azure OpenAI embeddings to find semantically similar documents in a corpus?

A BM25 keyword matching B Standard SQL full-text search queries C Cosine similarity between document embeddings and query embeddings using the text-embedding-ada-002 or text-embedding-3 models D TF-IDF vector comparison

49 / 60

Question 50 of 60

The Azure AI Engineer Associate AI-102 exam identifies which Azure AI Search feature as enabling document chunking and vectorization during the indexing process without external preprocessing?

A Manual chunking via Azure Functions before indexing B Integrated vectorization using a built-in skill that calls an Azure OpenAI embedding model during indexing C Azure Data Factory pipeline for vector generation D Azure AI Search BM25 indexing with synonym maps

50 / 60

Question 51 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Language feature enables a model to answer questions based on a provided text passage without pre-indexing?

A Conversational Language Understanding (CLU) B Question Answering with a single passage (extractive QA without a knowledge base) C Custom Text Classification D Named entity recognition

51 / 60

Question 52 of 60

According to AI-102 exam objectives, which Azure AI Language feature allows an organization to train a model that identifies proprietary entity types specific to their industry â€” such as equipment serial numbers, internal project codes, or regulatory clause references â€” that are not covered by built-in named entity recognition?

A Pre-built entity recognition using the standard NER model with no training data required B Custom named entity recognition (Custom NER) â€” a custom model trained on labeled examples to identify domain-specific entities such as product codes, contract clauses, or medical terms C Azure AI Translator custom glossary, which substitutes domain terminology during translation D Azure AI Speech custom speech model, which improves transcription accuracy for domain-specific vocabulary

52 / 60

Question 53 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure AI Speech feature transcribes conversations from multiple speakers and labels which speaker said what?

A Custom Speech acoustic model training B Conversation transcription with speaker diarization C Text-to-speech neural voice synthesis D Speech translation to another language

53 / 60

Question 54 of 60

On the Azure AI Engineer Associate AI-102 exam, which Azure Monitor feature enables setting up alerts when an Azure OpenAI endpoint's token consumption exceeds a defined threshold?

A Azure Policy compliance reports B Azure Monitor metric alerts on the Azure OpenAI 'Total Token Transactions' metric C Azure Advisor cost recommendations D Azure Security Center threat alerts

54 / 60

Question 55 of 60

The Azure AI Engineer Associate AI-102 exam covers which Azure AI Content Safety feature allows defining custom terms that should always be blocked in AI responses?

A Azure Policy custom definitions B Custom blocklists in Azure AI Content Safety C Azure Monitor alert rules D Azure OpenAI frequency penalty

55 / 60

Question 56 of 60

The Azure AI Engineer Associate AI-102 exam identifies which Azure AI Document Intelligence feature as extracting key-value pairs and table data from semi-structured documents?

A Read model (OCR only) B General Document model â€” extracts key-value pairs, entities, and tables from general documents C Custom classification model D Prebuilt Invoice model only

56 / 60

Question 57 of 60

According to the Azure AI Engineer Associate AI-102 exam, which AI-102 scenario MOST benefits from Azure AI Language's 'Conversational Language Understanding' (CLU) rather than QnA?

A Displaying pre-written FAQ answers based on keyword matching B Identifying user intent and extracting entities from natural language to trigger application logic (e.g., 'Book a flight to Paris on Friday') C Generating multi-paragraph explanations of complex topics D Extracting structured data from scanned invoices

57 / 60

Question 58 of 60

According to the Azure AI Engineer Associate AI-102 exam, which technique MOST reduces the latency of an Azure AI Search query for a high-traffic production application?

A Reducing the index size by deleting low-value documents B Enabling query result caching and using multiple search replicas for read scalability C Using only full-text search without semantic features D Increasing the search index partition count to the maximum

58 / 60

Question 59 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI Search tier is MOST appropriate for a production application requiring SLA guarantees, semantic search, and vector search?

A Free tier B Standard S1 or higher (with semantic search and vector field support) C Basic tier D Storage Optimized tier

59 / 60

Question 60 of 60

According to the Azure AI Engineer Associate AI-102 exam, which Azure AI service MOST directly enables voice biometric verification for a customer authentication scenario?

A Azure Custom Speech B Azure AI Speaker Recognition â€” Speaker Verification API C Azure AI Speech Translation D Azure Text Analytics

60 / 60