Chapter 3 of 7 · Databricks Context Engineer Associate
Knowledge Retrieval and Genie Configuration
Task context, RAG with AI Search, chunking, Genie curation and Unity Catalog governance.
- 4topics
- 21 minfast track
- 1diagrams and exhibits
- 20%of the exam
Topics in this chapter
Take them in order, or jump to the one you need.
- 1Task Context Elements and Retrieval TimingDatabricks Context Engineer: pick the context elements a task needs and choose pre-inference retrieval or just-in-time agentic…5 min
- 2RAG Pipelines, Chunking and AI Search ConfigurationDatabricks Context Engineer: build a governed RAG pipeline, choose a chunking strategy and fix the AI Search setting behind poor…Diagram6 min
- 3Unity Catalog Metadata and Genie Space CurationDatabricks Context Engineer: find the Unity Catalog metadata gap behind wrong agent answers and pick tables, views, queries and…5 min
- 4Retrieval Failure Modes and Authoritative Source GovernanceDatabricks Context Engineer: read MLflow eval logs and UC metadata to name a retrieval failure, then limit agent retrieval to…5 min
About this chapter
What this chapter covers
Knowledge Retrieval and Genie Configuration is 20% of the Databricks Context Engineer Associate exam, the largest chapter. It covers how an agent gets the right outside knowledge into its context: which facts a task needs, when to load them, how documents reach the model through Databricks AI Search, and how structured data reaches it through Genie spaces.
The chapter has four parts: choosing context elements and the retrieval timing (pre-inference or just-in-time), building a governed RAG pipeline with a sound chunking strategy and AI Search configuration, curating Unity Catalog metadata and Genie spaces, and diagnosing retrieval failures with MLflow and Unity Catalog governance.
Questions are single-answer and multiple-selection ("Which TWO ...?") questions. Almost all are scenarios that describe an agent, a trace, a configuration or MLflow evaluation results and ask for the diagnosis or the remediation that meets a stated constraint. There are no exhibits.
How the topics fit together
You first decide what the agent must know for the task and whether to load it before the model call or fetch it with a tool. Unstructured knowledge then flows through chunking and an AI Search index, structured knowledge flows through Unity Catalog metadata and Genie spaces, and MLflow plus Unity Catalog governance keep both channels accurate and limited to the right sources.
- Context element questions ask which instructions, data, tools or constraints a task cannot run without.
- Retrieval timing questions ask whether up-front loading or a retrieval tool call fits the use case.
- RAG and AI Search questions ask which chunking, index type, sync mode or query setting fixes poor retrieval.
- Genie and metadata questions ask which comment, instruction or trusted asset closes an accuracy gap.
- Governance questions ask which grant, filter or index scope keeps derived or stale data out of context.
Before you pick a fix, name the layer that failed. Stale text points to index sync, wrong chunks point to chunking or query settings, misread columns point to metadata, and an unofficial table in the trace points to governance. The distractors usually fix a different layer.
How to use these pages
Each topic page has the same shape, so you can skim for what you need.
- What it is: the idea in plain words.
- How it works: the mechanism, with tables and examples.
- What to remember for the exam: the facts to memorize and how questions are phrased.
- Common mistakes: the classic traps.
- Check yourself: three questions to test your understanding.
Ready for more practice?
The full Databricks Context Engineer Associate question bank has 600 questions, with timed practice exams and an explanation for every answer.
This study guide is independent and is not affiliated with, endorsed by, or sponsored by Databricks. Vendor names, certification names and exam codes are trademarks of their respective owners and are used for identification only.