Skip to content
All case studies
production2024 — 2025 · LightHouse Community Foundation

WeConnect 2.0

A three-role community platform — youth portal, admin dashboard, staff tools — built pixel-perfect from Figma and tuned for low-bandwidth field devices.

ReactTypeScriptZustandAxiosShadcn UIVitest

3

User roles served

0

Design defects reported

0

Milestones missed

The problem

A non-profit's field staff tracked youth programs on paper and disconnected tools. They needed one platform serving three very different users — youth members, program staff, and admins — that stays fast on cheap Android phones with patchy connectivity, and matches the funded design spec exactly.

Architecture

architecture · WeConnect 2.0
Youth PortalStaff DashboardAdmin DashboardShared Design Systemrole-aware componentsZustand Storeselective subscriptionsAxios Layerinterceptors · authREST APIserver-side pagination
  • Role-aware component architecture: one codebase renders three distinct experiences from a shared design system, with route-level code-splitting per role.
  • Dynamic data tables with multi-column sorting, filtering, and server-side pagination — the server does the heavy lifting so low-end devices never hold full datasets in memory.
  • Zustand for state with selective subscriptions: components re-render only on the slices they read, not on every store write.
  • Centralised Axios interceptors inject auth tokens and normalise error handling once, instead of per-request boilerplate across dozens of API modules.
  • React.memo + lazy-loaded routes keep the interactive bundle minimal; heavy admin views never ship to youth users.

Key engineering challenges

Fast on low-bandwidth field devices

The real users are field staff on budget phones. Server-side pagination, selective store subscriptions, and route-level lazy loading kept interactions responsive where a naive client-side table would have frozen the UI.

Zero design-drift across three roles

Funders signed off on the Figma. I maintained pixel-fidelity through rapid client-feedback iterations — the client reported zero design-to-implementation defects across all roles.

Where AI fits next

  • Natural-language report queries for program staff ('show attendance trends for the mentorship cohort').
  • Auto-summarised activity logs so admins read a digest, not raw entries.

Deployment

SPA served via CDN with environment-driven API endpoints; CI checks on every merge.