The complete Databricks Certified Generative AI Engineer Associate study library — 15 lessons mapped to every domain and topic in the official exam guide. Free to read, no account needed.
Create a free account to track completed lessons, bookmark topics and see your exam readiness. · Pro lessons unlock with a one-time upgrade.
Databricks publishes exactly what the Databricks GenAI tests and how much each part is worth. Here it is — every domain, its weight, the topics inside it, and the lessons that cover them.
Designing LLM applications: prompts that produce a required format, model tasks for a business need, chain components, multi-stage tool use and when to use Agent Bricks. Official weighting 14%.
Preparing knowledge for RAG: choosing source documents, extracting and filtering content, chunking strategies, writing chunks to Delta tables in Unity Catalog, evaluating retrieval and re-ranking. Official weighting 14%.
Building the application: frameworks such as LangChain, prompt augmentation and engineering, quality and safety assessment, guardrails, choosing LLMs and embedding models, and agentic systems with MLflow, Agent Framework and Genie. Official weighting 30%.
Assembling and deploying: pyfunc chains, RAG application elements, registering to Unity Catalog, Mosaic AI Vector Search, Model Serving and Foundation Model APIs, ai_query() batch inference, memory stores, CI/CD, prompt lifecycle, MCP servers and user interfaces. Official weighting 22%.
Governance for GenAI: masking and guardrails against malicious input, licensing of data sources, and mitigating problematic text in source data. Official weighting 8%.
Evaluating and monitoring: quantitative model selection, agent evaluation with MLflow scoring, tracing and custom scorers, judges and ground truth, SME feedback, inference tables, Agent Monitoring, AI Gateway and cost control. Official weighting 12%.
Reading every note front to back is the slowest way to pass. The exam is weighted, so your revision should be too — the blueprint above tells you exactly where the marks are.
Application Development carries the most marks on the Databricks GenAI, so it earns the most of your attention. Work down the blueprint from the heaviest domain to the lightest rather than starting at domain 1 and grinding forward.
Read each lesson once without stopping to memorise — you are building a map, not a memory. On the second pass, read only the "what's tested" and key-takeaway blocks. If a takeaway surprises you, that lesson goes back on the pile.
Cover a lesson's key takeaways and try to say them out loud from the heading alone. Recall you can produce under your own steam is the only recall that survives a timed exam; recognition on a re-read is not the same thing.
Finishing the notes is not the goal — scoring is. Move to Databricks GenAI practice questions early, then come back to the exact lessons your wrong answers point at. That loop is worth far more than a third read-through.
Exam structure, domain weights and scoring on this page come from the official Databricks GenAI exam guide.