Rohit Verma

I started in 2007 as a web designer cutting mockups and hand-writing table-less HTML, the era when getting a layout to survive Internet Explorer was a craft in itself. Design came first for me, engineering followed, and that order still shows: I think in interfaces, then build the machinery underneath them.

The years since have run through most of what the frontend became. Responsive builds for NASCAR, Harley-Davidson, and Sky at Sapient. Adobe Experience Manager portals for Honeywell. Around forty WordPress themes before that, which is a kind of apprenticeship nobody designs on purpose but everybody benefits from.

Nineteen years in, the thing I care about hasn't changed: software that people use without noticing the effort behind it. The stack rotates; the standard doesn't.

Since 2020 I've owned the frontend of Yuno Learning, a live-classes ed-tech platform, which is a different discipline from shipping a project and walking away. Booking, checkout, payments, instructor and admin consoles: surfaces where a regression costs real money the same afternoon.

Two builds from those years define how I work. Prasar, a white-label Flutter platform that ships more than ten branded apps (upGrad, RoboMonk, Shikshak Dwar among them) from a single codebase with automated build and branding pipelines. And the migration of the platform's SEO-critical pages to a Nuxt server-side-rendered application, with per-environment deploy pipelines, because search visibility is an architecture decision, not a marketing wish.

Alongside the code, I run the frontend team: hiring, mentoring, standards, and the unglamorous work of keeping a platform coherent while it grows.

I work AI-first, and I mean that as method, not marketing. In Cursor, projects I run carry written rules, reusable agent skills, and MCP integrations, so AI agents follow the same architecture the team does instead of inventing a new one every session. Design-to-code runs through Figma MCP: real components out of real design files. ChatGPT handles the thinking-out-loud layer: problem decomposition, documentation, communication drafts.

The judgment stays human. AI multiplies output; nineteen years of taste decides what ships. This site itself was built that way: agent-assisted, architecture-first, every pixel reviewed.