Context
Okiyumi grew from a single site into a group of independent products — storefront, rental platform, AI workspace, status center, private apps. Each product lives on its own stack and lifecycle, but they share one operational foundation.
Problem
Every new product needed the same things before it could be real: a domain, TLS, reverse proxy routing, container deployment, monitoring, backups. Rebuilding that per product would multiply cost and failure modes.
My role
I designed and built the shared infrastructure of the ecosystem and operate it in production: server infrastructure, reverse proxy and domain/TLS management, containerized deployment, monitoring and the operational conventions each service follows.
System
Independent services behind one ingress layer. Public traffic terminates at nginx with per-domain TLS; every service runs as a container on one host; monitoring and alerting are centralized; each product keeps its own data and release cycle.
Architecture
Host nginx → 127.0.0.1 service ports → containers. Certificates via Let's Encrypt. Health endpoints and a watchdog layer above the containers. Products communicate through explicit contracts, not shared databases.
Operations
Production operations are part of the design: a public status page, health checks, certificate renewal and backups run as routine, monitored processes rather than manual chores.
Current state
The ecosystem runs in production; new services are added by extending the same deployment and monitoring conventions. Not every line of every product is written by one person — the infrastructure, boundaries and operations are.
Facts
- One server runs the ecosystem's production deployment
- Reverse proxy, domains and TLS centrally managed
- Containerized deployment across all services
- 24/7 monitoring with a public status page
- Shared identity and operational conventions across products
- Service boundaries keep products independent