The question specifies that a query key (read-only) is compromised and explicitly states the requirement to "ensure that users only have read only access." Admin keys (Primary/Secondary) provide full read-write access and should not be used for client apps, as this violates the principle of least privilege.

To rotate a query key with zero downtime, you must follow this specific lifecycle:

1. Add a new query key: unlike Admin keys which are "regenerated," query keys are created as new items in a list (up to 50 per service).
2. Change the app to use the new key: Update the application configuration to direct traffic to the valid key.
3. Delete the compromised key: Once traffic has migrated, remove the old key to revoke unauthorized access.

Microsoft Learn - Connect to Azure AI Search using API keys:

"Query keys are used for read-only access to documents within an index... You can create up to 50 query keys per service." "The process for regenerating a query key is to delete and then recreate it."

Source: Manage Azure AI Search API Keys

Microsoft Learn - Azure CLI Command Reference (Query Keys):

"Unlike admin keys, query keys are not regenerated. The process for regenerating a query key is to delete and then recreate it."

Source: az search service query-key delete

Microsoft Learn - Security Best Practices:

"If you want to continue using API keys, be sure to always monitor who has access to your API keys and regenerate API keys on a regular cadence." (Context: Rotation involves overlap to prevent downtime).

Source: Security controls for Azure AI Search

To meet the requirement of displaying names and faces of people, the widgets parameter must include people. To ensure users can search for keywords, the standard configuration includes the keywords insight (to list them) and adds the search functionality via the controls parameter. Therefore, widgets=people,keywords and controls=search is the correct configuration.

For the Player widget, the requirement to "Show captions" dictates setting showcaptions to true. The requirement for the language to be "English (United States)" mandates setting the captions parameter to the culture code en-US.

Embed Azure AI Video Indexer widgets in your apps - Microsoft Learn

Reference: This document details the URL parameters for embedding widgets.

Section: "Insights widget" table defines widgets (options: people, keywords, etc.) and controls (options: search, etc.).

Section: "Player widget" table defines showCaptions (Boolean) and captions (Language code).

URL: https://learn.microsoft.com/en-us/azure/azure-video-indexer/video-indexer-embed-widgets

View closed captions - Azure AI Video Indexer - Microsoft Learn

Reference: Confirms the functionality of caption display and language selection in the player.

URL: https://learn.microsoft.com/en-us/azure/azure-video-indexer/view-closed-captions

To meet the requirement that managed feedback remains on the internal network, you must use Azure AI containers. This allows the processing of data to happen locally (on-premises) rather than sending the text payload to the public Azure cloud endpoints.

1. Provision the Language service resource: Even for local containers, you must first provision a resource in Azure. The container requires the API Key and Endpoint URL from this resource to authenticate and send billing/metering usage data back to Azure.
2. Deploy to on-premises server: You must pull the Docker image and place it on a server within your internal network (on-premises). Deploying to an Azure Container Instance would violate the data residency requirement by moving processing to the cloud.
3. Run and query: Finally, you instantiate (run) the container using the docker run command with the billing credentials. The application then queries the container's local endpoint (e.g., http://localhost:5000) for sentiment analysis, ensuring the feedback text never leaves the network.

Microsoft Learn. (2023). Install and run containers for Azure AI services. Retrieved from https://learn.microsoft.com/en-us/azure/ai-services/cognitive-services-container-support

Relevant Section: "Prerequisites" and "The host computer" (Confirming the need for an Azure resource and a local host computer).

Microsoft Learn. (2023). Install and run Docker containers for the Language service. Retrieved from https://learn.microsoft.com/en-us/azure/ai-services/language-service/sentiment-opinion-mining/how-to/use-containers

Relevant Section: "Run the container with docker run" and "Query the container's prediction endpoint."

Microsoft Learn. (2024). Azure AI services containers faq. Retrieved from https://learn.microsoft.com/en-us/azure/ai-services/cognitive-services-container-support#can-i-run-containers-in-a-disconnected-environment

Relevant Section: Highlights that while data processing is local, the container connects to the Azure resource for billing (unless specifically using disconnected containers, which still requires the initial resource setup).

The provided phrases are examples of user input and should be added as example utterances to the FindContact intent. Example utterances are the primary method for training a language model to recognize user intent by providing varied examples of what a user might say.

Patterns, on the other hand, are templates used to match specific text structures (e.g., "Search for {contact} in {location}"). They are useful for improving accuracy for well-defined formats but are not the correct tool for adding a diverse set of general training phrases. The provided phrases vary in structure and are better suited as individual example utterances for the model to learn from.

A. Yes: This is incorrect because patterns are not the primary or appropriate method for adding general training phrases. The correct approach is to add them as example utterances to the intent.

1. Microsoft Learn - Add example utterances in Language Studio: "Adding example utterances to your intents is a key part of your language understanding model. The model uses these utterances to learn how to predict the correct intents and entities." This document establishes that example utterances are the fundamental training data.

2. Microsoft Learn - Patterns in Conversational Language Understanding: "Patterns are used to find an exact match... Patterns are templates that should contain a mix of words and entities you want to match." This reference clarifies that patterns are for specific template matching, not for general training examples like the ones provided in the scenario.

3. Microsoft Learn - Intents in your Conversational Language Understanding model: "An intent represents a task or action the user wants to perform. It is a purpose or goal expressed in a user's utterance... You train your model by adding example utterances to each intent." This reinforces that the correct procedure is to add the phrases as example utterances to the intent.

The scenario requires identifying "unusual values" within 5,000 time series datasets generated by IoT sensors for predictive maintenance. The Azure Anomaly Detector service is specifically designed for this purpose. It provides a REST API that ingests time series data and uses machine learning models to identify anomalies, such as spikes, dips, and trend changes, without requiring prior machine learning expertise. This directly addresses the need to detect unusual patterns in sensor data to predict potential machinery failures.

B. Cognitive Search: This service (now Azure AI Search) is for indexing and querying large datasets to implement sophisticated search experiences, not for analyzing time series data for anomalies.

C. Form Recognizer: This service (now Azure AI Document Intelligence) is designed to extract text, key-value pairs, and tables from documents and forms, which is irrelevant to sensor data analysis.

D. Custom Vision: This is a computer vision service used to build custom models for image classification and object detection; it does not process numerical time series data.

1. Microsoft Learn. (2023). What is the Anomaly Detector API?. Azure AI services documentation. Retrieved from https://learn.microsoft.com/en-us/azure/ai-services/anomaly-detector/overview.

Reference Point: The "Overview" section explicitly states, "The Anomaly Detector API enables you to monitor and detect abnormalities in your time series data... Use the Anomaly Detector API to detect anomalies in your business data, or from IoT devices." This directly maps the service to the question's scenario.

2. Microsoft Learn. (2023). Quickstart: Detect anomalies in your time series data using the Anomaly Detector REST API and Python. Azure AI services documentation. Retrieved from https://learn.microsoft.com/en-us/azure/ai-services/anomaly-detector/quickstarts/client-libraries.

Reference Point: This quickstart provides a practical example of using the service with a JSON file containing time series data points, demonstrating its core function which aligns with the problem statement.

3. Microsoft Learn. (2024). What is Azure AI Search?. Azure AI Search documentation. Retrieved from https://learn.microsoft.com/en-us/azure/search/search-what-is-azure-search.

Reference Point: The "Overview" section describes the service as a "fully managed search as a service for web, mobile, and enterprise apps," confirming it is not for time series anomaly detection.

4. Microsoft Learn. (2024). What is Azure AI Document Intelligence?. Azure AI Document Intelligence documentation. Retrieved from https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/overview.

Reference Point: The first paragraph states its purpose is to "extract text and structure from documents," which is unrelated to the IoT sensor data in the question.

To configure the performance of an Azure OpenAI model deployment using an ARM template, you must set the capacity property within the sku object. This property defines the model's throughput in thousands of Tokens-Per-Minute (kTPM).

Azure OpenAI Service enforces a rate limit ratio of 6 Requests-Per-Minute (RPM) for every 1,000 Tokens-Per-Minute (TPM). To meet the requirement of 600 RPM, you must calculate the corresponding TPM:

1. Calculate required TPM: Required TPM=6Required RPM​×1,000 Required TPM=6600​×1,000=100×1,000=100,000 TPM
2. Convert TPM to the capacity value: The capacity property is measured in units of 1,000 TPM. capacity=1,000100,000 TPM​=100

Therefore, setting the capacity to 100 correctly provisions the resource to handle 600 requests per minute.

Microsoft Learn, Azure OpenAI Service quotas and limits. This document details the relationship between Tokens-Per-Minute (TPM) and Requests-Per-Minute (RPM). In the section "Managing quota," it states, "For standard deployments the requests-per-minute (RPM) limit is proportional to the tokens-per-minute (TPM) limit assigned... the ratio between TPM and RPM is 6 RPM per 1,000 TPM."

Microsoft Docs, Bicep & ARM template reference for Microsoft.CognitiveServices/accounts/deployments. The official template reference specifies the valid properties for the sku object of a deployment. It lists name (string) and capacity (integer) as the key properties for defining the deployment's performance tier and throughput.

The Microsoft responsible AI principle of inclusiveness aims to create AI systems that empower everyone by being accessible to people with a wide range of abilities and backgrounds. Adding the Direct Line Speech channel extends the bot with voice-in and voice-out capabilities. This enhancement makes the bot accessible to users with visual impairments, motor disabilities, or those who simply prefer voice interaction over typing. By providing an alternative, more accessible interaction modality, the bot becomes more inclusive, directly adhering to the specified principle.

A. Implement authentication for bot1: This addresses the principles of security and privacy by verifying user identity, not the principle of inclusiveness which focuses on accessibility.

B. Enable active learning for Iu1: This is a technique to improve the language model's accuracy over time, aligning with the principle of reliability and safety, not inclusiveness.

C. Host Iu1 in a container: This is a deployment strategy for managing data locality or running in disconnected environments; it does not affect user accessibility or inclusiveness.

1. Microsoft Learn, "The six principles for responsible AI": Under the "Inclusiveness" section, it states, "AI systems should empower everyone and engage people. When we design these systems, we need to address a broad range of human needs and experiences. Inclusive design practices can help developers understand and address potential barriers that could unintentionally exclude people." Adding a voice interface is a direct application of this principle.

Source: Microsoft Learn, Module: "Introduction to responsible AI", Unit: "What are the guiding principles for responsible AI?".

2. Microsoft Learn, "About Direct Line Speech": "Direct Line Speech is a robust, end-to-end solution for creating a flexible, extensible, voice assistant... It's powered by the Bot Framework and its Direct Line Speech channel, which is optimized for voice-in, voice-out interaction with bots." This documentation confirms that Direct Line Speech is the specific technology for adding voice accessibility.

Source: Microsoft Learn, Azure Bot Service Documentation, "Direct Line Speech overview".

3. Microsoft Learn, "Add authentication to a bot": "Often a bot must access secured resources on behalf of the user... To do this the bot must be authorized to access these resources based on the user's credentials." This confirms authentication is for security and authorization, not inclusiveness.

Source: Microsoft Learn, Azure Bot Service Documentation, "Add authentication to a bot in Bot Framework SDK".

4. Microsoft Learn, "Active learning": "Active learning is the process of selecting the new utterances that are unclear and adding them to your training data. This helps the model become more accurate." This defines active learning as a model improvement technique, related to reliability.

Source: Microsoft Learn, Azure AI services Documentation, "Active learning".

The document processing requirement states that a manual review is necessary if a key field's confidence score is less than 70 percent (0.7). The provided JSON output from the Azure AI Document Intelligence (formerly Form Recognizer) API shows the extracted MerchantName field has a confidence score of 0.65. The expression documentResults.fields.MerchantName.confidence < 0.7 correctly evaluates this condition, as 0.65 is less than 0.7. This expression accurately targets the specific field with low confidence, thereby triggering the required manual review process as per the requirements.

A. This expression only verifies that the document is a receipt; it does not evaluate any confidence scores to trigger a manual review.

B. This expression is syntactically invalid. It does not correctly reference the confidence property of any specific field within the fields object.

C. This expression uses the incorrect comparison operator (>). It checks for high confidence, which is the opposite of the requirement to flag low-confidence results for review.

1. Microsoft Azure AI Document Intelligence Documentation, "Prebuilt receipt model": This document outlines the JSON structure returned by the prebuilt receipt model. It confirms that each extracted field, such as MerchantName, has an associated confidence property, which is a numeric value between 0 and 1. The structure documentResults.fields.MerchantName.confidence is the correct path to access this value.

Source: Microsoft Learn, Azure AI services documentation.

Reference: https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/concept-receipt?view=doc-intel-4.0.0, see the "Data extraction" section for field schema.

2. Microsoft Azure AI Document Intelligence Documentation, "Model quality and confidence scores": This official documentation explains the purpose of confidence scores and recommends their use for validation. It states, "The confidence score indicates the probability that the extracted value is correct... For scenarios that are sensitive to accuracy, we recommend that you use a human-in-the-loop review." This directly supports the question's premise of using a confidence threshold (70%) to trigger a manual review.

Source: Microsoft Learn, Azure AI services documentation.

Reference: https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/concept-model-overview?view=doc-intel-4.0.0#model-quality-and-confidence-scores

3. Microsoft Azure AI Document Intelligence SDK Quickstart: The official SDK quickstarts provide code examples demonstrating how to access the analysis results. The code consistently shows accessing a specific field's confidence score by navigating through the result object to the specific field name, such as document.Fields["MerchantName"].Confidence (in C#), which corresponds to the JSON path in the question.

Source: Microsoft Learn, Azure AI services documentation.

Reference: https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/quickstarts/get-started-sdks-rest-api?view=doc-intel-4.0.0&pivots=programming-language-csharp, see the section "Analyze a document from a URI".

To deploy an Azure Custom Vision model to a mobile device (Android), the model must be exported to run locally. Standard domains, such as the "Retail" domain mentioned in the scenario, are cloud-only and do not support export.

1. Change the model domain: You must switch the project settings from the standard "Retail" domain to a "Compact" domain (e.g., "General (compact)" or specific object detection compact domains) to enable export capabilities.
2. Retrain the model: Changing the domain alters the underlying neural network architecture (optimizing it for edge devices). Consequently, the model must be retrained to adapt to this new architecture.
3. Export the model: Once trained on a compact domain, the "Export" button becomes active, allowing you to download the model in formats like TensorFlow (for Android), CoreML, or ONNX.

Microsoft Learn. "Export your model for use with mobile devices - Custom Vision Service." Azure AI services. Accessed November 20, 2025.

Relevant Section: "Convert to a compact domain" - "To export a model, you must ensure that you've selected a compact domain... The standard domains are not exportable."

Relevant Section: "Export your model" - Describes the process of downloading the files after training on a compact domain.

URL: https://learn.microsoft.com/en-us/azure/ai-services/custom-vision-service/export-your-model

Microsoft Learn. "Select a domain for a Custom Vision project." Azure AI services. Accessed November 20, 2025.

Relevant Section: "Compact domains" - Explains that compact domains are optimized for real-time constraints on mobile devices and are required for export.

URL: https://learn.microsoft.com/en-us/azure/ai-services/custom-vision-service/select-domain

The cURL command is being constructed to call the Azure AI Content Safety REST API.

1. The first selected option, contentsafety, is required because the API endpoint for the Azure AI Content Safety service specifically uses this path segment to route requests to the correct service within the Cognitive Services resource.
2. The second selected option, /image:analyze, specifies the exact REST API operation to be performed. This operation is designed to analyze an image for various types of harmful content, including violence, hate, self-harm, and sexual content, which directly addresses the user's need to validate the resource's capabilities.

The other options are incorrect. contentmoderator refers to a legacy service. vision refers to the separate Azure AI Vision service. /detect and /ProcessImage are not valid operations for this API.

Microsoft Azure Documentation, "Quickstart: Azure AI Content Safety": This guide provides sample code, including cURL commands, for using the service. The examples explicitly show the endpoint structure "{endpoint}/contentsafety/image:analyze?api-version=....

Reference: In the "Analyze image content with the REST API" section, the sample cURL command demonstrates the required URL format.

Microsoft Azure Documentation, "Analyze Image - REST API (Azure AI Content Safety)": The official API reference documentation confirms the HTTP POST request structure for image analysis.

Reference: The "Request URL" section specifies the endpoint as POST {Endpoint}/contentsafety/image:analyze. The "Request Body" section also details how to reference an image via a blobUrl.