Work
Mira Auto — AI parts verification
Live at mi-ra.rf: VIN → factory OEM → honest city prices → anti-counterfeit checklist. No in-house cart, no invented part numbers.
- Next.js
- TypeScript
- Postgres
- DeepSeek
- Zzap
Project screenshots
Task
Build a service that stops car owners from buying fakes, wrong crosses, and overpriced retail. Enter a VIN, OEM, or part name — get the factory number from the catalog, a city price cut, and a receive-time authenticity checklist. If the catalog is silent, the product says miss instead of inventing a “close enough” SKU.
Constraints
One engineer, no separate design/content team. No checkout — deep-links to seller stock only. Never invent OEM without catalog proof. Prices via partner APIs, no silent storefront scraping. Ship on mi-ra.rf, not as a Figma prototype.
Shipped
Production on mi-ra.rf: landing with VIN/OEM/part search and car chips; traps block and Exist/Zzap comparison; 100+ public teardown cards; OEM card with Moscow-vs-city price and anti-fake checklist; garage dashboard; API selection stream; light/dark theme; solid mobile width.
Result
Not a brochure — a working product: in a minute the visitor gets the offer, opens a teardown or runs a query, then leaves via deep-link with no service markup. Mobile holds. The same shape fits another niche: storefront + cabinet + integrations where the system answer earns trust instead of a closed tab.