"This course contains the use of artificial intelligence."
Turn Databricks AI Playground experiments into evaluated GenAI prototypes you can defend and productionize.
Getting an LLM to return one impressive answer is easy. Building a prompt that remains useful against changing inputs, enterprise data, quality requirements, and production constraints is a different problem.
This practical course gives you a repeatable workflow for designing prompts in Databricks AI Playground, evaluating outputs systematically, comparing models and parameters, connecting prototype decisions to governed data workflows, understanding usage and cost considerations, and preparing the winning experiment for production integration.
Who this course is for:

Data engineers building GenAI features on Databricks.

Data scientists and ML engineers adding LLM workflows to their skill set.

AI and application engineers evaluating model endpoints.

Analytics engineers prototyping natural-language applications.

Technical architects and product professionals evaluating Databricks GenAI use cases.
What you will learn:

Design reliable system and user prompts.

Convert business requirements into measurable LLM instructions.

Create representative evaluation datasets.

Compare prompt and model candidates systematically.

Test supported generation parameters through controlled experiments.

Design prompts around enterprise data and grounding requirements.

Recognize hallucination, instruction, and output-format failures.

Evaluate quality alongside usage, latency, and estimated cost.

Translate a Playground experiment into a production handoff.

Build an employer-ready Databricks GenAI portfolio project.
Requirements:

Access to a Databricks workspace with the required AI/serving capabilities enabled by your administrator.

Access to at least one suitable model or serving endpoint.

Basic familiarity with Databricks.

Basic Python or SQL is useful for production exercises but is not required for the core prompt-engineering lessons.

Use synthetic course data unless you are authorized to process organizational data.
Final project:
You will build an enterprise support intelligence prototype in Databricks AI Playground. You will compare at least three prompt/model configurations against a minimum 10-case evaluation set, score them against five criteria, incorporate grounded business context, document generation and usage/cost assumptions, select a winner, and produce a production handoff package.

More Info