Conceptual Layer

Conceptual Layer

I help European platform teams implement Dagster, migrate job-centric pipelines to assets, and operate Dagster on Kubernetes. My current work focuses on the orchestration practices needed to keep analytics, AI, and agentic workloads supplied with observable data. I also created the Orchestration Maturity Model, a framework for assessing how a platform represents and operates its data.

Work with me on Dagster
Level 1Ad HocScripts & cronNo alertingZero lineageManual runsReference onlyLevel 2ScheduledCentral schedulerFailure alertsJob-centric modelManual dependenciesCommon starting pointLevel 3OrchestratedAsset-awareLineage-nativeFull CI/CDDefined re-run scopeChecks & metadataAsset modelLevel 4ObservableFreshness SLAsCross-system lineageAutomated remediationService metricsLevel 5PlatformData contractsTraceable metricsSupported self-serviceComposable pipelinesLevel 3 adds an explicit model of data assets.conceptuallayer.com
The five levels of data orchestration maturity

Latest Articles

article10 min read

The Data Orchestration Maturity Model: A Field Guide

A reference guide for assessing whether your orchestration platform manages scheduled jobs, data assets, or a shared data product.

orchestrationmaturity modeldata platformsdagster
article14 min read

The Ceiling of Job-Centric Data Platforms

A job scheduler can report that code ran. Data teams eventually need to know whether the data is ready, correct, and safe to rebuild.

orchestrationdata platformsdagstermaturity model

From the Blog

blog6 min read

The LLM Loop: A Mental Model for Better AI Work

A practical loop for improving LLM work by changing the instructions, context, model, or next iteration after each answer.

llmsai-assisted documentationdata engineering

Stay in the loop

New writing on orchestration, data platforms, and the Orchestration Maturity Model, delivered to your inbox.

Subscribe on Substack