Talk decks

The PDFs from my talks on Airflow, dbt and self-service analytics, free to keep. Each one leaves out the animation steps, so it reads page by page.

A full room watching Ismail presentIsmail on stage with a microphone
Slide titled A Data Platform to Enable Teams to Go Faster, drawn by hand
A company in chaos, 15 teams with 15 ways to use AirflowFreedom for teams against discipline by the data platformMigrating from Airflow 2 to 3 by changing one package versionIsmail presenting this deck to a seated audience

Taking back control and scale with Airflow

In English, 38 slides

How a platform team took 15 teams from 15 ways of using Airflow to one package.

  • The technical debt behind 15 ways of using Airflow
  • Why docs, templates and support didn't fix it
  • The Python package that moved teams to one YAML file
  • The controls that keep hundreds of workflows in line
Download the PDF: Taking back control and scale with Airflow, 4.3 MB
Slide titled We're aligned, we ship better data, with Alice, Thomas and Ismail drawn by hand
We stopped helping people choose, we removed the choicedbt is an engine, but an engine alone can't move a carA tool without rules is difficult to manage

My feedback on self-service analytics

In English, 58 slides

Alice, Thomas and Ismail share one platform and learn to ship data on their own.

  • Why more docs and more support didn't scale
  • How scoring every dbt project made the standards stick
  • The Airflow setup that lets analytics engineers focus on SQL
  • The before and after figures, from onboarding to support
Download the PDF: My feedback on self-service analytics, 6.1 MB
Slide asking whether a new engineer could understand your whole platform in a day, with a dog at a laptop
The labyrinth, four orchestrators and no visibilityThe hamster wheel of fixing the same problems againWhat changed with Airflow, before and afterIsmail presenting this deck, with the slide on platforms becoming unmanageable behind him

How we unified data orchestration for 80+ engineers

Deck in French, 20 slides

Four orchestrators, no visibility, and how Airflow became the one place to look.

  • The visible problem and the invisible one
  • What changed for Alice and Thomas with Airflow
  • DAG templates, custom operators and standard environments
  • How the platform reached 200+ DAGs in production
Download the PDF: How we unified data orchestration for 80+ engineers, 1.8 MB