Get Your CertGenAIPlayCompete
‹ GenAI

Databricks Certified Generative AI Engineer Associate

1 test · 10 questions
Browse all 10 questions ›
Your progress
Test 110 questions

Study guides

AI engineer roadmap 2026: skills, certs and the orderAI engineer roadmap for 2026: skills to learn in order, AI engineer vs ML engineer, which certifications fit each stage, and what to skip.
NVIDIA AI certifications 2026: GenAI, agentic and infraNVIDIA AI certifications in 2026: generative AI, agentic AI and infrastructure exams, with codes, levels, prices, formats and first choices.
Databricks Generative AI Engineer Associate exam guideWhat the Databricks Generative AI Engineer Associate exam covers: format, US$200 price, domains, free Fundamentals badge and preparation.
Best AI certifications 2026: AWS, Google, Microsoft, ClaudeThe best AI certifications in 2026 from AWS, Google, Microsoft, Claude and Databricks, by level, with price, status and which one to take first.
What Is RAG? How Retrieval-Augmented Generation WorksWhat RAG is and how it works: chunking, embeddings, vector search, hybrid search, reranking, GraphRAG and agentic RAG on AWS, Google and Azure.
Prompt engineering vs RAG vs fine-tuningWhen to use prompt engineering, retrieval-augmented generation, fine-tuning or continued pre-training, compared on cost, freshness, hallucination and privacy.

About the Databricks Generative AI Engineer Associate exam

The Databricks Certified Generative AI Engineer Associate exam is for engineers who build applications on top of large language models inside the Databricks platform: retrieval-augmented generation (RAG) apps, agents that call tools, and batch jobs that run a model over a table. It checks that you can take a business problem, break it into a chain or agent, prepare the data it needs, serve it, and then govern and measure it once it is live. Databricks suggests at least six months of hands-on work with the tasks in the exam guide before you sit it.

The questions are mostly short scenarios. A team has a RAG app that returns stale answers, an agent that needs to call a SQL function on behalf of the user, or a chatbot that leaks personal data, and you choose the Databricks feature or design step that fixes it. Wrong options tend to be things that would help in general but not with the stated problem, or that take more work than the platform feature built for the job.

Exam format

  • Questions: 45 scored multiple-choice questions. The exam may also include unscored items, with extra time allowed for them.
  • Time: 90 minutes.
  • Cost: 200 USD.
  • Delivery: Proctored, online or at a test center. No test aids are allowed.
  • Languages: English, Japanese, Portuguese (Brazil) and Korean.
  • Validity: 2 years. To recertify you take the current version of the exam again.

Databricks does not publish a passing score on the exam page.

What the exam covers

The exam outline has six sections:

  • Design Applications (14%). Turning a requirement into a chain or agent: which model tasks are needed, what the prompt should return, and when a ready-made option such as Agent Bricks fits better than custom code.
  • Data Preparation (14%). Chunking documents to suit the embedding model's limits, cleaning and storing chunks in Delta tables, and keeping a Vector Search index in step with its source table through Delta Sync and change data feed.
  • Application Development (30%). The largest section: retrieval choices such as hybrid search for exact terms like part numbers, prompt design, tool calling, and diagnosing poor or hallucinated answers using MLflow traces.
  • Assembling and Deploying Apps (22%). Logging and registering a chain or agent with MLflow and Unity Catalog, versioning prompts with the prompt registry and aliases, serving through Model Serving, authentication for the resources an agent uses, and batch inference with ai_query.
  • Governance (8%). Guardrails in AI Gateway such as PII filtering, access control through Unity Catalog, and keeping sensitive data out of prompts and responses.
  • Evaluation and Monitoring (12%). MLflow evaluation with LLM judges, knowing which judges need ground-truth answers and which do not, and tracking quality once an app is in production.

How the practice questions here are organised

There is one 10-question practice test for this exam at the moment. It touches all six sections, with the extra weight on application development and deployment that the real exam has, so it works as a quick check of whether your knowledge is broad enough rather than as full preparation. Each question explains why the right answer is right and why each other option is wrong, with a link to the Databricks documentation it is based on. More tests will be added over time.

A study plan that works

  • Take the practice test first, without studying, to see which sections are weak.
  • Build one small RAG app end to end in a Databricks workspace: chunk a few documents, create a Vector Search index, log the chain with MLflow, serve it and query it. Most exam scenarios are about one step of that flow going wrong.
  • Read the explanation of every question you missed or guessed. The notes on the wrong options often name the feature you didn't know existed.
  • Spend extra time on MLflow tracing and evaluation. Knowing what a trace shows and which judge to use comes up across several sections.
  • Learn the product names as they are now. Databricks renames and replaces features often, and the exam follows the current documentation, not older blog posts or courses.

These questions are original, written from the public Databricks documentation and the official exam guide. They are not real exam questions, and this site is not affiliated with Databricks.