10 lessons aligned with the official exam guide of the new Databricks Generative AI Engineer Associate certification. From LLM fundamentals to RAG, Vector Search, Agent Framework, Agent Bricks and MLflow Tracing — all from a production engineer's lens.
End-to-end: from understanding LLMs to deploying and governing agents in production with the Databricks AI stack.
Prompt engineering, context engineering, when to fine-tune, model selection.
Document parsing, chunking, embeddings, Databricks Vector Search, hybrid retrieval.
Code-first agents with LangGraph, declarative agents with Agent Bricks, when to use each.
Observability for AI apps, prompt versioning, evaluation sets, prod telemetry.
Tool calling with Unity Catalog Functions, Genie spaces, Model Context Protocol.
Model Serving, Databricks Apps, Review App, monitoring, cost and quality KPIs.
Same order as the official exam guide. First-mover course in EN/PT.
What the exam tests, the Databricks GenAI stack at a glance, decisions you make in real projects.
How LLMs work for engineers, prompt vs context engineering, when to fine-tune vs when to retrieve.
Real-world parsing (PDF, HTML, text), chunking strategies, metadata enrichment, the dataset behind a RAG.
Databricks Vector Search end-to-end, hybrid retrieval, filters, ANN indexes, evaluation of retrieval quality.
Foundation Model APIs, AI Playground for exploration, AI Functions for SQL batch, when to use each.
Tracing AI apps in production, prompt versioning, evaluation sets, building feedback loops.
Tool calling with Unity Catalog Functions, Genie spaces for natural language analytics, Model Context Protocol.
Building agents with LangGraph + Mosaic AI Agent Framework, multi-turn, multi-tool, multi-agent patterns.
Declarative agents with Agent Bricks, human-in-the-loop with Review App, calibration workflows.
Model Serving endpoints, Databricks Apps, cost and quality monitoring, exam strategy and full mock.
GenAI Engineer Associate is part of our 6-course family in the Gold Plan — alongside Associate, Professional, PySpark Free, SDP and DP-750.
Not deep ML, but you should be comfortable with Python and Databricks basics. If you have the Associate already, you're set. We focus on the engineering side of GenAI, not on training models from scratch.
Recommended yes — Foundation Model APIs and Vector Search have limits on Free Edition. We provide setup guides for both paths.
Two ways: (1) it maps to the actual Databricks certification, with mock questions in the exam format; (2) we teach the Databricks stack specifically — Vector Search, Agent Framework, Agent Bricks, AI Gateway — not the Python ecosystem in general.
RAG is one of the 10 lessons. The rest covers agents (code-first and declarative), tools, evaluation, deployment, governance. RAG is necessary but not sufficient for the exam.
Databricks launched it in 2025. Our course has been ready since June 2026. We're one of the first courses to cover it in both English and Portuguese.