Modernizing without the big-bang rewrite

DTU's digital platform supported approximately 30 active subsites on a large, tightly coupled legacy codebase.

Modernization was necessary, but replacing the whole platform at once would have introduced significant technical and business risk. Any change to shared infrastructure could affect the sites already running on it, which made a traditional rewrite both difficult to validate and expensive if something went wrong.

The goal was to move DTU toward a new design system and a more modern frontend architecture without disrupting the sites already running on the platform.

A big-bang rewrite would have meant rebuilding and validating a substantial amount of functionality before any of the modernization could safely reach production — and with a shared platform underneath the active subsites, the blast radius of getting it wrong was significant.

I introduced an approach where legacy and modern components could coexist: replacing parts of the platform progressively rather than committing to a high-risk cutover. This follows the same principle commonly associated with the Strangler Fig Pattern — incrementally replacing a legacy system rather than rewriting it all at once.

I introduced an incremental modernization strategy and the design system underneath it, structured so that new components could stand in for legacy ones without breaking anything running in production.

Rather than replacing existing components immediately, new versions were developed and validated alongside them. This let us migrate component by component, keep the active subsites operating throughout, and limit the blast radius of any given change to the specific piece we were replacing.

I also introduced a GitFlow-based development and release process to create clearer control around how changes moved through development and into production. With a shared platform underneath a live footprint, release strategy was part of the architecture. Modernization needed to happen without making everyday development and releases unpredictable.

DTU gained a practical path from a tightly coupled legacy platform toward a modern design system without committing to a high-risk rewrite.

The design system and frontend architecture were robust enough that legacy components could be replaced with new ones incrementally — piece by piece, in production, without disrupting the active subsites depending on the platform.

Instead of a single project with a hard cutover, modernization became an incremental technical strategy that could evolve alongside the existing platform.