Platform architecture at scale
How two engineers sustain eleven applications.
The problem
A growing portfolio, a two person frontend team, and every new application re-solving authentication, HTTP, UI, configuration and CI from scratch.
The decision
A single workspace with a shared package layer, so a new application inherits single sign on, typed backend access and quality gates on day one.
The result
Seven of eight applications scaffolded onto that platform, three of them within a single fortnight.
The governance
Contract driven integration: backend interface specifications are version controlled and types are generated rather than hand written, so a contract change breaks the build instead of production.
The durability
Written operating rules that let AI agents and backend engineers work safely in the workspace, plus a handover guide and a contract map.
What I would do differently
Two real tradeoffs stand in the workspace today. Two UI systems coexist, which means two ways of building the same screen. And test coverage is strong in some areas and thin in others.