Databricks Certified Generative AI Engineer Associate Exam Questions

Databricks Certified Generative AI Engineer Associate Exam Objectives

1. 1

   1.0 Design Applications

   • 1.1Designing a prompt that elicits a specifically formatted response is the focal point of this topic. It also covers selecting model tasks to accomplish a given business requirement. Moreover, the topic focuses on selecting chain components for a desired model input and output.
2. 2

   2.0 Data Preparation

   • 2.1Sub-topics are about applying a chunking strategy for a given document structure and model constraints. The topic covers filtering extraneous content in source documents that degrades quality of a RAG application. Sub-topic about choosing the appropriate Python package to extract document content from provided source data and format is also included. Moreover, the topic defines operations and sequence to write given chunked text into Delta Lake tables in Unity Catalog.
3. 3

   3.0 Application Development

   • 3.1It covers creating tools needed to extract data for a given data retrieval need. Selecting Langchain/similar tools for use in a Generative AI application is also discussed. The topic focuses on identifying how prompt formats can change model outputs and results. Qualitatively assessment of responses to identify common issues such as quality and safety is also addressed here. Selecting chunking strategy based on model & retrieval evaluation is discussed as well. Moreover, the topic focuses on augmentation of a prompt with additional context from a user's input based on key fields, terms, and intents.
4. 4

   4.0 Assembling and Deploying Applications

   • 4.1It has sub-topic about coding a chain using a pyfunc model with pre- and post-processing. The topic covers controlling access to resources from model serving endpoints. Moreover, coding a simple chain according to requirements is also discussed here.
5. 5

   5.0 Governance

   • 5.1Usage of masking techniques as guard rails to meet a performance objective is discussed in this topic. It also covers selecting guardrail techniques to protect against malicious user inputs to a Gen AI application. Lastly, the topic focuses on recommending an alternative for problematic text mitigation in a data source feeding a RAG application.
6. 6

   6.0 Evaluation and Monitoring

   • 6.1Selecting an LLM choice (size and architecture) based on a set of quantitative evaluation metrics is the focal point here. This topic also covers selecting key metrics to monitor for a specific LLM deployment scenario. Lastly, the topic covers evaluation of model performance in a RAG application using MLflow.