Task 1.3: Implement Data Validation and Processing Pipelines for FM Consumption. AIP-C01 Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A financial services company is using a SageMaker AI endpoint hosting a fine-tuned Llama 3 model for a document summarization task. The application receives documents in PDF format, extracts the text, and sends it to the endpoint. The team has noticed that the model's responses are inconsistent and sometimes contain hallucinated information. They want to improve the quality of the input data and the consistency of the model's output. What should they do?

   • AUse SageMaker Data Wrangler to create a visual flow that samples the documents, identifies missing values, and generates a data quality report.
   • BUse Amazon Bedrock to reformat the extracted text into a more structured format, such as JSON, before sending it to the SageMaker endpoint.
   • CUse Amazon Comprehend to detect the language of the document and translate it to English before sending it to the model.
   • DUse a Lambda function to clean the extracted text, remove any non-UTF-8 characters, and truncate the input to the model's maximum context length before invoking the endpoint.
   Reveal Solution Discussion Select an answer first
2. 7foundation · easy

   A machine learning engineer wants to visually inspect and prepare a tabular dataset for foundation model training. The engineer needs to identify missing values, outliers, and class imbalances, and then apply transformations before exporting the prepared data. Which AWS service provides a visual interface for these data preparation steps without writing code?

   • AAWS Glue DataBrew
   • BAmazon SageMaker Data Wrangler
   • CAmazon QuickSight
   • DAWS Glue Studio
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   A developer is building a chatbot that uses a foundation model. The model expects a conversation history to be formatted as a list of messages, each with a role (e.g., 'user' or 'assistant') and content. Which type of formatting is this?

   • AConversation formatting
   • BJSON formatting
   • CPrompt engineering
   • DStructured data preparation
   Reveal Solution Discussion Select an answer first
5. 9application · medium

   A healthcare company is building a Bedrock-powered clinical documentation assistant. They receive unstructured clinical notes from multiple hospitals in JSON files. The data quality team wants to enforce a rule that every note must contain a valid patient ID, a non-empty 'note_text' field, and a date in ISO 8601 format before the data is used for FM inference. The pipeline currently uses AWS Glue to catalog the data and a Lambda function to call Bedrock. What is the MOST operationally efficient way to implement these validation rules?

   • AUse Amazon Comprehend to detect the entities in the note text and verify that a patient ID and date are present in the extracted entities.
   • BCreate an AWS Glue Data Quality rule set with rules for completeness, data type, and custom SQL checks, and run it as part of the Glue job before the Lambda invocation.
   • CWrite a custom Python script in the Lambda function that parses each JSON record and raises an exception if any field is missing or malformed.
   • DConfigure a SageMaker Data Wrangler flow to import the JSON, apply a custom transformation to validate the fields, and export the cleaned data to S3.
   Reveal Solution Discussion Select an answer first
6. 10foundation · easy

   A developer wants to improve the quality of text input before sending it to a foundation model. The developer needs to identify and extract key entities (such as names, dates, and locations) from the text so that this information can be used to enrich the prompt. Which AWS service can be used for entity extraction?

   • AAmazon Textract
   • BAmazon Comprehend
   • CAmazon Rekognition
   • DAmazon Transcribe
   Reveal Solution Discussion Select an answer first