Where the engineering habits started: owning an EMS end-to-end at a seed-stage startup.

01 · Context

RaSoft is where I started — first as a Frontend Engineer Intern in December 2018, then as a Frontend Engineer through February 2021. It was seed-stage, which in practice means there is no separate frontend team, no design system, and no one else to hand the hard part to. Two problems sat in front of me. The company's own website was underperforming: the UI was dated, it wasn't responsive, and people dropped out before they ever made contact. And internally, employment management had no system at all — it needed to be designed and built.

02 · The decision

I took end-to-end ownership of the UI rather than picking up tickets. On the site, I revamped the interface around responsive layout and accessibility, on the theory that most of the lost conversions were friction rather than persuasion. On the EMS, I delivered the whole thing from concept to production in React.js, JavaScript, HTML, and SCSS — designing the interactions with the backend engineers rather than waiting for a spec. In a startup that small, the choice on every feature is speed versus a structure that survives the next six months, and I made that call consciously each time instead of always defaulting to speed.

03 · The trade-off I made

Rapid delivery vs. scalable architecture. The honest answer is that I leaned slightly toward architecture and shipped a little slower than I could have — refactoring components rather than duplicating them, and improving performance at the component level while features were still landing. At seed stage that's a defensible but not free choice: some things reached users later than they might have. What I'd defend is the direction, because the EMS was going to be lived in daily, and the site was the company's front door.

04 · What I measured

The site had a number: UX and conversion improved ~15% after the responsive redesign and accessibility work — reported internally, over the months after launch, and it was one of several things changing at once. The EMS was the other half of the job: an internal system that did not exist, designed and built end to end.

05 · Outcome

A complete Employment Management System delivered from concept to production with end-to-end UI development, and a company site rebuilt for responsiveness and accessibility that lifted conversion ~15%. Along the way I ran code reviews and pair-programming sessions to hold quality, improved performance through refactoring and component-level optimization.

06 · What I'd do differently

What I'd do differently: I optimized components before I had any measurement, which is guessing with extra steps. Now I'd profile first and let the numbers pick the target — a habit that later turned into the 40% and 15% load-time wins at Kodez and Axinom, both of which started from a measurement rather than an instinct. The lesson I carry from RaSoft: pair programming and code review were where I actually learned to write maintainable code, and both cost time that always felt like it could be spent shipping. It couldn't.