Case study · Stanverse Technologies · Aug 2025 – Present

Making stanshop.co faster and easier to buy from

stanshop.co is a creator commerce platform processing 30Cr+ in GMV every month. As a Software Developer there, I worked on the storefront's performance, its purchase flow and the architecture underneath both.

Next.jsReact NativeTypeScriptSSRBFF patternMonorepoCore Web Vitals

Buy conversion

30% → 60%

5–6×

Performance score

10 → 50–60

−45 pts

Poor LCP on Android

~80% → ~35%

30Cr+

Monthly GMV served

Production scale

Before → after

Figures marked ~ are approximate.

Before After

Buy conversion

higher is better

0
50
100
30%
Before: 30%
60%
After: 60%
BeforeAfter

Next.js performance score

higher is better

5–6×
0
50
100
10
Before: 10
50–60
After: 50–60
BeforeAfter

Poor LCP · Android

lower is better

−45 pts
0
50
100
80%
Before: 80%
35%
After: 35%
BeforeAfter

Poor LCP · mobile web

lower is better

−40 pts
0
50
100
75%
Before: 75%
35%
After: 35%
BeforeAfter
View as table
MetricBeforeAfterChange
Buy conversion30%60%
Next.js performance score1050–605–6×
Poor LCP · Android80%35%−45 pts
Poor LCP · mobile web75%35%−40 pts

What I did

Performance overhaul

Took the Next.js performance score from 10 to 50–60 with code splitting, lazy loading and server-side rendering. This cut poor LCP from ~80% to ~35% on Android and from ~75% to ~35% on mobile web, and lifted good LCP to ~25–30%.

Navigation & buy flow

Optimized the navigation UX flows, which doubled buy conversion from 30% to 60%.

BFF + config-driven UI

Architected a backend-for-frontend layer that serves UI configuration, so features can be rolled out dynamically without shipping a new build.

Monorepo architecture

Designed a monorepo for Stanshop and Photon, improving developer productivity and CI/CD efficiency.

Partner whitelabeling

Built Partner Tag whitelabeling, so multiple partners run their own branded storefronts from a single codebase.

Fliq trading integration

Integrated the Fliq blockchain trading platform via WebView, supporting ~500K transactions.