Legacy platform migration
A legacy platform moved onto Next.js, TypeScript and Node.js one route at a time, with the old stack still serving everything that had not been migrated yet.
- Role
- Full stack - architecture to deployment
- Stack
- Next.js, TypeScript, Node.js, AWS
- Timeline
- 2025 - 2026
- Client
- MiFin Services Inc.
01
The problem
The platform ran on a stack that had stopped paying for itself. Every change meant reading around the previous workaround first, and the areas the business wanted to move fastest were the ones most expensive to touch. Stopping to rewrite it would have meant months without new features.
02
The approach
Migrate route by route instead of rewriting the platform. The old application kept serving whatever had not moved, each migrated route shipped on its own, and the work stayed reversible: a route that misbehaved went back rather than taking a release with it.
03
The build
Next.js and TypeScript on the front, Node services behind it, and typed contracts at every boundary between the new code and what was left of the old, so a shape that changed on one side failed at build rather than in production. Deployed on AWS on the same pipeline as the rest of the platform.
Outcome
- 0
- Feature freezes during the migration
- 2
- Stacks in production until the last route moved
- 1
- Codebase and deployment path at the end