AI-103 Practice Test

Answer choices

1. A. Azure Cosmos DB with custom indexing policies and vector search capabilities, within the documented scope, ownership, and validation boundaries.
2. B. Azure OpenAI Service with custom embeddings and a retrieval-augmented generation (RAG) pipeline, as the primary proposed approach.
3. C. Azure AI Search with vector search capabilities and schema-based filtering, as the recommended implementation across the complete governed service lifecycle.
4. D. Azure Machine Learning with a custom vector similarity model deployed as a web service, in context.

Wrong-answer review

• A. Azure Cosmos DB with custom indexing policies and vector search capabilities, within the documented scope, ownership, and validation boundaries.: Cosmos DB indexing can support operational data patterns, but it is not the managed search design requested here.
• B. Azure OpenAI Service with custom embeddings and a retrieval-augmented generation (RAG) pipeline, as the primary proposed approach.: Azure OpenAI can generate embeddings, but it does not by itself provide the filterable search index.
• D. Azure Machine Learning with a custom vector similarity model deployed as a web service, in context.: A custom ML web service would require building search and filtering behavior outside the managed search service.

Answer choices

1. A. Pass the dimensions parameter set to 512 in the API request body when calling the embeddings endpoint
2. B. Configure the deployment in Azure OpenAI Studio to crop the output dimension size, for the required operational result and control objective.
3. C. Set the temperature parameter to 0.1 to limit the vector size, for the stated implementation and support requirements.
4. D. Divide the output vector array by 6 in your client application code, for the required operational result and control objective.

Answer choices

1. A. The API will ignore the unsupported parameters and process the request successfully using the model's default settings, for the described technical objective and its associated operational control requirements.
2. B. The API will return a 400 Validation Error, because reasoning models require temperature, top_p, and presence_penalty to remain at their default values and setting them otherwise is unsupported, within organization-wide risk-and-accountability boundaries.
3. C. The API will process the parameters normally, dynamically scaling the reasoning token length based on the temperature, for the described technical objective and its associated operational control requirements, for review.
4. D. The API will return a 401 Unauthorized Error, since custom parameters are only allowed on dedicated provisioned throughput units deployments, for the described technical objective and its associated operational control requirements.

Answer choices

1. A. Create a custom function tool in Python and code the HTTP request manually in the agent's run loop, as selected.
2. B. Configure an OpenAPI tool or plugin in the agent's tool definition, importing the OpenAPI JSON schema directly, within the defined security and accountability boundaries.
3. C. Configure the Azure AI Search indexer tool to crawl the external REST endpoint, within the documented scope, ownership, and validation boundaries.
4. D. Deploy a custom Azure AI Search skillset that executes the API calls during document ingestion, within the defined security and accountability boundaries.

Answer choices

1. A. Call the begin_analyze_document method, retrieve the operation ID, and poll get_analyze_document_result until the operation succeeds, as the recommended implementation across the complete governed service lifecycle.
2. B. Send a synchronous HTTP POST request to the endpoint and retrieve the extracted table and text data directly in the response body, for the described technical objective and its associated operational control requirements.
3. C. Call the analyze_document_sync method, passing the file stream, and wait for the thread to block until the analysis finishes, as the primary proposed approach.
4. D. Use the AnalyzeDocument method to download the layout model locally, perform the extraction in the application container, and upload the output JSON

Answer choices

1. A. Configure an Azure Event Grid subscription that calls the search API directly on every database write, for the specified implementation requirement.
2. B. Configure a custom skillset that queries the Cosmos DB change feed in real-time, under the documented operational and governance requirements.
3. C. Enable public network access on the search service, which enables real-time synchronization by default, within the documented scope, ownership, and validation boundaries.
4. D. Configure a schedule using the schedule property in the indexer definition to run the indexer at regular, automated intervals

Answer choices

1. A. Optimize the chunk size and embedding model used for vector search to improve the accuracy of retrieved knowledge base documents, within the documented operational, security, ownership, and validation requirements.
2. B. Increase the content filter sensitivity threshold to prevent the model from generating potentially inaccurate responses, as the proposed design for the complete governed operational workflow.
3. C. Adjust the prompt template used in the Azure OpenAI service to explicitly instruct the model to only respond based on the provided knowledge base, within organization-wide risk-and-accountability boundaries.
4. D. Modify the evaluation flow in Azure AI Foundry to prioritize relevance over coherence, ensuring the chatbot's responses directly address the user's query, within the described operational context.

Answer choices

1. A. Azure AI Search with vector search capabilities, for the stated requirement.
2. B. Azure Cosmos DB with vector search capabilities, for this scenario.
3. C. Azure SQL Database with a custom vector similarity index, under the stated decision criteria.
4. D. Azure Data Lake Storage Gen2 with a custom search application, within this context.

Answer choices

1. A. A tool function that returns a placeholder order-status string without sending an HTTP request or authentication header, under the documented operational and governance requirements.
2. B. A tool function that sends an HTTP POST request with an Authorization bearer token header and a JSON body containing orderId
3. C. A tool function that sends an HTTP GET request with an X-Auth-Token header and order_id as a query parameter, for this requirement.
4. D. A tool function that sends an HTTP POST request with a Basic authentication-style header and form-encoded orderId data, as the organization’s selected response.

Answer choices

1. A. Modify the user prompt to include detailed instructions about the desired persona and data privacy guidelines for each interaction, for the stated security, delivery, and accountability requirements.
2. B. Implement a complex chain of prompts using Azure AI Prompt Flow to dynamically adjust the chatbot's behavior based on user input, within cross-functional operational-accountability boundaries.
3. C. Utilize a fine-tuned Azure OpenAI model specifically trained on a dataset of customer service interactions that exemplify the desired persona and data handling practices, for the described technical objective.
4. D. Create a custom system prompt that defines the chatbot's persona, outlines data privacy protocols, and provides examples of appropriate responses, in the described situation.