The current chapter is Calvinball Technology in Singapore, where the work has moved decisively into AI-native product design. I own the frontend architecture on a data-to-insight-to-action SaaS platform, which means I am responsible not just for screen-level implementation, but for how the whole product stays coherent as the surface area expands.

One of the biggest builds so far has been Studio, an internal and client-delivery tool that lets teams configure the product at a granular level without touching application code. That changed the deployment model. New client setup time dropped from 40 days to 14, a 65% reduction, and delivery no longer required product-level code edits for every new implementation.

I also built the full agentic chat feature from scratch. That included designing the backend SSE tunnel and rebuilding the frontend from a static interface into a live streaming experience with progressive rendering and motion, closer in feel to tools like ChatGPT or Claude than to a standard enterprise message box.

The role has also required deeper security ownership than any earlier chapter. For a client-mandated external audit, I handled the VAPT compliance implementation personally: CSP headers, input sanitisation, and auth hardening across the platform. It was a good reminder that frontend architecture is not separate from security posture. In modern products, they are often the same conversation.

On top of that, I redesigned the product UI to make it feel more alive and responsive. Users consistently described the new interface as more engaging than the old flat version. That mattered because AI products are judged as much by interaction quality as by raw model output. If the interface feels dead, the product feels weaker than it is.

Calvinball feels like a convergence point for everything that came before: system thinking from the full stack years, performance discipline from commerce and restaurant tech, and the ability to step into an imperfect codebase and make it legible again. The difference now is that the product problems are more abstract, and the pace of change is much higher.