Your Node or React SaaS was built in 2014. It still earns money. The original team has moved on. I keep it running, patch the security holes, ship new features, and slowly drag it toward something maintainable. Production AWS environments, the 3am bugs, releases that don't break the build. Most engagements: ongoing, 1–3 days a week.
I've migrated production frontends from Angular to React/TypeScript twice, on systems that couldn't go down — including a public-facing SaaS used by structural engineers worldwide. The trick is gradual: new features ship in React inside the old shell, route by route, component by component. No big-bang rewrites, no two-year transitions, no broken users. Most engagements start with an audit and turn into a 6–12 month gradual migration.
I started with Java in university, fell into Ruby in 2008, and got pulled into the open-source DataObjects project before pull requests existed — you had to be granted commit access. I wrote JDBC adapters for the JRuby version, finished my M.Sc. on the same project, and helped run Eurucamp in 2012.
Since then I've been the long-running engineer on a family of production SaaS apps in the structural BIM industry — first prototyping a public 3D-model sharing platform in Node with MongoDB/GridFS, later refactoring it off that onto third-party storage. Then building a licensing & organisation management tool from scratch. Both frontends migrated from Angular to React/TypeScript over the years, gradually, in production, without rewrites. None of these systems have been rewritten end-to-end. They've been kept alive and made better while the business kept billing customers.
I've worked remotely the whole time, mostly through Finnish prime contractors. I deployed with Capistrano before there was Heroku. Today I'm exploring how to retrofit AI features (Claude, OpenAI, agentic flows) into long-lived production systems — the part most teams will need next, and the part nobody has a clean playbook for yet.
