I inherited a working backend and built the entire front end on top of it — every screen designed and shipped against real, live production data, not mocks.
Built on GitHub in a dedicated design branch, then merged straight into production.
An AI property-pricing platform for the Hungarian market, now live in the US too — 200,000+ historical transactions, live data sources, and a machine-learning model behind every estimate. One address in, a market valuation in under ten seconds out.
One platform, four audiences — homeowners, brokers, developers, banks — each with its own use case on the same data.
The full flow in nine screens — homepage search, audience tabs, the valuation wizard, the report, pricing — cut hard, no fades.
A three-step wizard collects the property's basics, building type, and condition before the model runs.
Every report carries a market value, three pricing scenarios by how fast you want to sell, a map, comparable sales, and a price-trend chart — plus AI-generated listing descriptions and a PDF export.
Free to start — no card, an instant result, three valuations a month, GDPR-compliant by default.
React front end on the inherited Django backend — Leaflet and MapLibre for the maps, Recharts for the trend charts, and Framer Motion for the animated diagrams — all code-driven, not recordings. Hungarian and English throughout via i18next. Designed and coded by me, on the live product.