Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 

Repository files navigation

Knowledge sharing - Declarative Data Pipelines

Table of Content (ToC)

Created by gh-md-toc

Overview

This project aims at collecting and sharing reference material about declarative data pipelines, that is, what they are, how to design and to build them.

That goes without saying that AI agents/assistants will be used wherever relevant. For instance, to begin with, a starting point/cornerstone is the Databricks AI Dev Kit and related material:

Even though the members of the GitHub organization may be employed by some companies, they speak on their personal behalf and do not represent these companies.

References

Data Engineering helpers

AI helpers skill sets

Articles and projects of interest

Spec-driven development (SDD)

Databricks declarative development

Deploy ETLs using Lakeflow Declarative Pipelines

Spec-driven workflow

Lakehouse Plumber

  • Overview:

The Metadata Driven framework for Databricks Lakeflow Declarative Pipelines (LDP) (formerly Delta Live Tables (DLT)). Metadata framework that generates production ready Pyspark code for Lakeflow Declarative Pipelines

Metadata-Driven Lakehouse Ingestion

In my initial experiments to democratize onboarding in metadata-driven development (MDD) architectures, the process has been much easier with Databricks AI Dev Kit.

I first built the infra foundation with DAB + MDD, then integrated AI Dev Kit and tested it at Databricks app level.

Technically, what we added:

  • Chat-based onboarding entrypoint
  • AI Agent runtime for intent-driven orchestration
  • MCP layer for Databricks tool execution
  • Lakebase for conversation and project-state persistence
  • Git-based metadata flow (YAML + DAB)
  • CI/CD gates for validation and controlled deployment

Impact so far:

  • Faster source/table onboarding
  • More consistent metadata-driven pipelines
  • Clear path from request -> config -> validation -> deployment

Also, development can be managed via CLI + Claude I tested Codex and Gemini Pro 3.1 as well; for this workflow Claude 4.6 performed best in my experience.

About

Knowledge sharing - Material about declarative data pipelines

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors