Databricks Certified Generative AI Engineer Associate - Prep

Discover how generative AI uses deep learning and large language models to generate content, and how context engineering with RAG, chunking strategies, tool calling, and AI agents enhances responses.

Explore interactive prompt engineering in the AI Playground using Databricks-hosted foundation models (GPT, Claude, Lama) and models served via Databricks model serving, adjust temperature, and apply system prompts.

Explore in-context learning through prompt engineering in a hands-on lab. Use context to ground model responses, limit hallucinations, and leverage internal knowledge bases, with a RAG workflow coming next.

Design a rag system with an llm for answers, an embedding model, a vector store, and a retrieval engine, plus optional re-ranker and prompt augmentation for accurate responses.

Advance the medallion data flow by refining bronze to silver for rag, parsing binary documents into text with ai_parse_document, and converting json to clean markdown via ai_query for silver_docs_parsed.

Compare chains and ai agents in rag implementation, highlighting retrieval needs and token efficiency. Learn how a retrieval agent uses vector search and an llm as tools to answer queries.

Develop a customer service chatbot for inquiries. Handle purchases, route questions, perform vector store retrieval, check inventory, and generate a payment link with an ai agent using Unity Catalog functions.

An AI customer service agent uses Unity Catalog functions and a vector store to handle product inquiries and purchases via inventory checks and checkout URL generation.

Evaluate generative AI models using MLflow 3's genai.evaluate with an evaluation dataset, a prediction function, and MLflow scorers to measure retrieval and answer quality against ground truth.

Evaluate generative ai models with MLFlow GenAI by defining a predict function from a Unity Catalog model and assessing with evaluation dataset and scorers such as relevance, safety, and fluency.

Deploy Gen AI models with Mosaic AI model serving by creating a serverless endpoint, selecting version 2 and CPU compute, and enabling batch and real-time inference.

Explore Agent Bricks, a no-code declarative framework for building production-grade AI agents, with a focus on the Knowledge Assistant using retrieval augmented generation, vector indices, and human-in-the-loop feedback.