
Complex multi-role real estate platform — buyers, sellers, agents, and investors — with live MLS integration, a real-time share auction system, and gated membership. Full product ownership from information architecture through full-stack delivery.

Three simultaneous user types — buyers, sellers, agents — with different information needs, different trust levels, and different decision timelines, all inside one platform. Added complexity: live MLS data integration, a real-time share auction system with bidding timers, gated membership tiers, Stripe-powered payments, and legal compliance requirements. The challenge wasn't just building features; it was designing a coherent product experience that didn't collapse under competing stakeholder needs.

Ran stakeholder sessions to surface conflicting priorities early, then used information architecture to separate each user type's critical path before writing a line of code. Built the design system in Figma first — component library, user flow maps for buyer/seller/agent/investor journeys, and visual identity with the blue/green gradient aesthetic. Evaluated stack options against two constraints: pre-revenue speed and the real functional ceiling of composed tooling. Chose Webflow + Xano + Wized for frontend speed and backend scalability, with Memberstack for access control and Stripe for payments.

Responsive platform with personalized dashboards per user role, dynamic MLS listing search with filter and detail pages, real-time share auction system with bidding timers and winner selection logic, Stripe-integrated membership tiers and ZIP Code products, and a full beta launch with built-in feedback instrumentation. Owned frontend, backend, and API integration — including custom Xano requests to third-party MLS data sources.

Launched to market with strong beta feedback. The platform worked as designed — the business wound down for non-technical reasons. What the project demonstrated: no-code/low-code tooling has a real ceiling for complex stateful products (real-time bidding, multi-role access, live third-party data), and exactly where that ceiling sits. This constraint directly informs how I now evaluate stack choices for AI product work — where composed tools fit and where custom infrastructure is the only honest answer.