BHyve v1 was one frontend talking to twelve separate backend repos, each with its own chart and deploy. I started the new web app and became the largest contributor to Platform v2: one Nx monorepo with 14 apps, 24 shared libraries and one release process.
A pages-router Next.js frontend where I wrote 671 of 693 commits, plus 12 NestJS services in separate repos with separate deploys.
One monorepo: a Next.js 16 web app, an iOS app, a Python AI service and 11 more apps, sharing 24 libraries and one CI.
The problem
Every feature touched three or four repos, every deploy was its own ritual, and the frontend was built on a router and a set of patterns that had reached their limit.
The idea
Rebuild on one monorepo with shared types and one release process, and design the new interface as a system first so every screen is built from the same parts.
How I did it
- Created the new web app on 11 July 2025 and wrote 727 of its 1,036 commits.
- Moved 42 components from the design prototype into the app (Dec 2025), then grew the library to 64, with flat, shadow-free styling and 33 per-screen design audits.
- Started the Python AI service (bhyve-ai) and the iOS app inside the monorepo, and owned the AI answer module (130 of its 165 commits).
- Cut the first production tag on 28 May 2026, the same day the new CI/CD landed.
The impact
The product now ships from one place, about 1.7 releases a day. My side alone is around 300,000 lines added to the new frontend and 174,000 to the platform API. The old frontend went quiet after January 2026 and saw its last commit in September.
A rewrite is only worth it if it changes how the team works. The design system and the single release process mattered more than the framework upgrade.