Skip to content

AI and LLMs

This lab explores practical generative AI and data science patterns in Fabric, including Azure OpenAI integration, LangChain workflows, document-oriented analysis, and MLflow-backed model tracking.

flowchart LR D[Fabric data] --> N[Notebook] N --> AOAI[Azure OpenAI] N --> LC[LangChain workflow] N --> ML[MLflow experiment] AOAI --> O[Grounded output] LC --> O ML --> O

Learning objectives

  • Configure Azure OpenAI connectivity for a Fabric-oriented workflow.
  • Build prompt templates and processing chains with LangChain.
  • Apply large-scale document/literature-review patterns to extract useful content.
  • Train, track, compare, and register machine-learning models with MLflow.

Lab notebook

The Fabric LLM overview sample notebook contains the runnable example. Work through it only after configuring the required Fabric, Azure OpenAI, identity, and network access settings described in the AI and LLMs guide.

Warning

Do not send sensitive or regulated data to external model endpoints without an approved data-handling design. Validate identities, tenant settings, access controls, regional availability, and model usage costs.

Suggested validation

  1. Confirm the notebook can authenticate without embedding credentials.
  2. Test prompts with non-sensitive sample data before connecting production sources.
  3. Review generated outputs for grounding, accuracy, and safety requirements.
  4. Record experiment parameters and evaluation evidence with the model artifact.