Databricks' exam outline for DBX-GENAI as of March 18, 2026, mapped to the chapters of Ultra Transcenders Databricks Generative AI Engineer Associate.
| Objective | Chapters |
|---|---|
| Section 1: Design Applications (14%) | |
| Design a prompt that elicits a specifically formatted response | 2. Prompt design and refinement |
| Select model tasks to accomplish a given business requirement | 1. From business need to LLM solution |
| Select chain components for a desired model input and output | 1. From business need to LLM solution |
| Translate business use case goals into a description of the desired inputs and outputs for the AI pipeline | 1. From business need to LLM solution |
| Define and order tools that gather knowledge or take actions for multi-stage reasoning | 3. Designing agents and using Agent Bricks |
| Determine how and when to use Agent Bricks (Knowledge Assistant, Multiagent Supervisor, Information Extraction) to solve problems | 3. Designing agents and using Agent Bricks |
| Section 2: Data Preparation (14%) | |
| Apply a chunking strategy for a given document structure and model constraints | 5. Chunking strategies and embedding context length |
| Filter extraneous content in source documents that degrades quality of a RAG application | 4. Finding and extracting source documents |
| Choose the appropriate Python package to extract document content from provided source data and format | 4. Finding and extracting source documents |
| Define operations and sequence to write given chunked text into Delta Lake tables in Unity Catalog | 6. Writing chunks to Unity Catalog and building an index |
| Identify needed source documents that provide necessary knowledge and quality for a given RAG application | 4. Finding and extracting source documents |
| Use tools and metrics to evaluate retrieval performance | 7. Tuning retrieval: configuration, metrics and re-ranking |
| Design retrieval systems using advanced chunking strategies | 5. Chunking strategies and embedding context length |
| Explain the role of re-ranking in the information retrieval process | 7. Tuning retrieval: configuration, metrics and re-ranking |
| Section 3: Application Development (30%) | |
| Select Langchain/similar tools for use in a Generative AI application | 9. Coding chains and pyfunc models |
| Qualitatively assess responses to identify common issues such as quality and safety | 13. Guardrails and legal risk |
| Select chunking strategy based on model & retrieval evaluation | 5. Chunking strategies and embedding context length |
| Augment a prompt with additional context from a user's input based on key fields, terms, and intents | 2. Prompt design and refinement |
| Create a prompt that adjusts an LLM's response from a baseline to a desired output | 2. Prompt design and refinement |
| Implement LLM guardrails to prevent negative outcomes | 13. Guardrails and legal risk |
| Select the best LLM based on the attributes of the application to be developed | 8. Selecting models |
| Select an embedding model context length based on source documents, expected queries, and optimization strategy | 5. Chunking strategies and embedding context length |
| Select a model from a model hub or marketplace for a task based on model metadata/model cards | 8. Selecting models |
| Select the best model for a given task based on common metrics generated in experiments | 8. Selecting models |
| Utilize MLflow and Agent Framework for developing agentic systems | 10. Agent systems: Agent Framework, Genie, MCP and memory |
| Compare the evaluation and monitoring phases of the Gen AI application life cycle | 14. Evaluating agents with MLflow |
| Enable multi-agent systems to leverage Genie Spaces or conversational API to retrieve data | 10. Agent systems: Agent Framework, Genie, MCP and memory |
| Section 4: Assembling and Deploying Applications (22%) | |
| Code a chain using a pyfunc model with pre- and post-processing | 9. Coding chains and pyfunc models |
| Control access to resources from model serving endpoints | 11. Logging, registering and serving |
| Code a simple chain according to requirements | 9. Coding chains and pyfunc models |
| Choose the basic elements needed to create a RAG application: model flavor, embedding model, retriever, dependencies, input examples, model signature | 11. Logging, registering and serving |
| Register the model to Unity Catalog using MLflow | 11. Logging, registering and serving |
| Create and query a Vector Search index | 6. Writing chunks to Unity Catalog and building an index |
| Identify how to serve an LLM application that leverages Foundation Model APIs | 11. Logging, registering and serving |
| Explain the key concepts and components of Mosaic AI Vector Search | 6. Writing chunks to Unity Catalog and building an index |
| Identify batch inference workloads and apply ai_query() appropriately | 11. Logging, registering and serving |
| Configure vector search for a particular solution based on number of embeddings, update frequency, latency, and cost requirements | 7. Tuning retrieval: configuration, metrics and re-ranking |
| Configure a persistent datastore to store and retrieve intermediate memory or structured information | 10. Agent systems: Agent Framework, Genie, MCP and memory |
| Apply CI/CD best practices such as updating a Vector Search index, promoting prompts across environments, and testing individual components of an agent | 12. Prompts, CI/CD and user interfaces |
| Integrate managed, external, and custom MCP servers based on a given application requirements | 10. Agent systems: Agent Framework, Genie, MCP and memory |
| Apply prompt version control and manage prompt lifecycle | 12. Prompts, CI/CD and user interfaces |
| Develop an appropriate interactive user facing interface for an agent usage scenario (Apps, Slack, Teams, etc.) | 12. Prompts, CI/CD and user interfaces |
| Section 5: Governance (8%) | |
| Use masking techniques as guard rails to meet a performance objective | 13. Guardrails and legal risk |
| Select guardrail techniques to protect against malicious user inputs to a Gen AI application | 13. Guardrails and legal risk |
| Use legal/licensing requirements for data sources to avoid legal risk | 13. Guardrails and legal risk |
| Recommend an alternative for problematic text mitigation in a data source feeding a GenAI application | 13. Guardrails and legal risk |
| Section 6: Evaluation and Monitoring (12%) | |
| Select an LLM choice (size and architecture) based on a set of quantitative evaluation metrics | 8. Selecting models |
| Select key metrics to monitor for a specific LLM deployment scenario | 15. Monitoring and controlling cost in production |
| Evaluate agent performance using MLflow scoring and tracing | 14. Evaluating agents with MLflow |
| Use inference logging to assess deployed RAG application performance | 15. Monitoring and controlling cost in production |
| Use Databricks features to control LLM costs | 15. Monitoring and controlling cost in production |
| Use inference tables and Agent Monitoring to track a live LLM endpoint | 15. Monitoring and controlling cost in production |
| Identify evaluation judges that require ground truth | 14. Evaluating agents with MLflow |
| Use AI Gateway (Inference Tables, Usage Tables, and rate limiting) to track an LLM or agent deployed via Agent Framework | 15. Monitoring and controlling cost in production |
| Use Databricks custom Scorers for evaluating agents and LLMs | 14. Evaluating agents with MLflow |
| Incorporate SME feedback to improve agent performance | 14. Evaluating agents with MLflow |
The full list of tasks under each objective is in the official DBX-GENAI exam guide. The book is organised by technology, so one chapter often serves several objectives.