LogeMoi Web — Original Real Estate Marketplace.

Real estate search in Bénin needed a proper tag-based discovery platform, built from scratch: a public website, separate from the admin panel and the backend API, where searching should work the way people actually describe a place they want — typing a partial word like 'pis' and being offered the real tag 'Piscine' — rather than free-text search against unstructured listing descriptions.

Full marketplace, not just listings. Built the public web app end to end on Next.js (App Router): tag-based property search, property detail pages, an announcer-facing dashboard (listings, billing, performance, prospects, verification), and a tenant account area (favorites, saved search alerts, a comparison tool with shareable links, subscription management).

Real payments and real maps, not stubs. Integrated FedaPay for checkout, including its own return-bridge pages, and Google Maps/Places autocomplete for location search and geocoding — the same patterns later carried over into the mobile app's own FedaPay integration.

Iterating the design in place. Shipped three successive home-page and search-page generations (home-v1/v2/v3, search-v2/v3) as the product direction evolved, rather than treating the first design as final — the current live site runs on the third generation.

Challenge — recognizing the architecture no longer fit the content. As the catalog increasingly came from short, vertical TikTok-sourced video rather than manually written listings, a desktop-era grid/list browsing model stopped matching how that content actually reads. Rather than retrofit a mobile-first video feed onto an existing desktop-first codebase, the right call was to scope and build a separate app for it instead.

Challenge — deciding what NOT to carry forward. The mobile-first rebuild deliberately dropped features rather than porting all of them: in-app messaging and generic notifications were cut for its first version, phone+OTP login was replaced with email/password and Google sign-in (OTP carried real cost risk from SMS-pumping/OTP-bombing abuse), and cross-app single sign-on between the two was explicitly scoped out once this app was already slated for retirement — shipping with two separate logins was the pragmatic choice over building shared auth for a soon-to-be-retired app.

In hindsight: the most valuable part of this project turned out to be recognizing its own limits — knowing when a working, feature-complete app is still the wrong foundation for where the product needs to go next, and having the discipline to scope a clean rebuild instead of layering a mismatched UI on top.