
Bink started as a startup-competition concept, and I turned it into a working prototype: a shopping assistant with conversational, multimodal search. You can describe what you’re after, or upload a photo.
Google Gemini drives the conversation. It understands text and image input, makes fashion recommendations, and generates the search keywords. SerpApi then pulls real-time results from Google Shopping, with a general mode and a “thrift” mode for second-hand finds. It also has a virtual try-on.
I built the full stack: a React frontend deployed on Vercel, a Node.js backend on Render, and Supabase for user authentication and API rate limiting.
Built with
- React
- Node.js
- TypeScript
- Gemini API
- SerpApi
- Supabase
- Vercel
- Render