Data org design for the AI era
Team topologies that work when analytics, ML, and platform blur together — and the ones that don't.
The old org chart — a data engineering team, an analytics team, a data science team, and a squabble over who owns ML — no longer maps to how work gets done.
The topology that consistently ships in 2026 has three groups. A platform group that owns the substrate: warehouse, orchestration, observability, semantic layer. A domain group that owns the business-facing work: dashboards, models, and copilots for a specific business unit. And an enablement group that owns tooling, standards, and the developer experience across both.
Analytics engineers and ML engineers live in the domain group. Platform engineers live in the platform group. Analytics leaders and ML leaders report to a single head of data. This is not an org-chart preference — it is a routing decision that determines how fast decisions get made.
The topology that keeps failing: a centralized data team that services every business unit through tickets. It scales to about 80 people and then collapses under its own coordination cost.
James advises boards and CDOs on data strategy. Former CDO at a global bank and a top-3 retailer.
More from JamesRelated articles
How to measure your analytics team's impact
A no-vanity-metric framework for showing the business what your data team is worth.
Hiring analytics engineers who will still be great in year two
A structured interview loop that filters for the traits that actually predict long-term impact.
Data contracts are boring — and that's the point
Why the least glamorous idea in the modern data stack is quietly the highest-leverage one for teams tired of 3AM pages.