I like owning an app end to end: the architecture, the backend, the UI, the deploy pipeline and the dashboards that tell you it's working. When there's no designer, I'll do the design pass too.
Most of the job is deciding what to build, what to leave out and which boring option will actually hold up. I use AI coding agents for a lot of the typing, behind tests and code review, but I know how everything I ship works and why it was built that way. The case studies here are written around those decisions.
I'm comfortable landing in codebases I've never seen. At Arbisoft that has meant a large Go fork of Apache DevLake, an Odoo ERP, a Next.js audio archive and a Webflow build, picking up each stack as I went. The approach stays the same: read the existing system, write a plan with its non goals, check the assumptions against real data and then build the simplest thing that meets the requirement. The authentication and telemetry work for DevLake both changed direction once the product needs or the data said so.
Before Arbisoft I spent about a year and a half at vFairs looking after 60+ production scrapers and building internal data tools, test automation and event AI features. I did my CS degree at NUST, where my final year project was Finasaan.