2025
Dragon Candy
An AI-powered marketplace that gets restaurants mouth-watering food photos and reels in hours, not weeks.
- Dragon Candy · dragoncandy.io
- AI full-stack engineer
- ongoing
- React · TypeScript · Supabase · shadcn/ui · Stripe · OpenAI · Tailwind
The problem
Restaurants and cafes need a steady stream of professional food photography and video to stay visible on social — but traditional production is slow, expensive, and full of back-and-forth. Briefing a shoot, finding the right creator, and turning content around can take weeks, which is fatal for time-sensitive promotions.
The approach
I built Dragon Candy as a two-sided marketplace in React with an AI assistant, “Donny,” at its center. Donny reads a restaurant's website to draft a campaign brief, then matches it to vetted creators scored on style, audience, and engagement. The platform handles the full loop — briefs, creator matching, booking, delivery, and a rush “DragonDash” lane for same-day turnarounds — built on Supabase for data and auth, Stripe for payments and payouts, and a shadcn/ui interface, with LLM steps for brief generation and structured matching. Creator and restaurant dashboards share primitives so both sides stay fast and consistent.
The outcome
Restaurants go from a website URL to a ready-to-shoot brief and a matched creator in minutes, and content lands in hours instead of weeks. The AI layer removes the blank-page problem on both sides — operators don't have to write briefs, and creators get clear, on-brand direction — while the marketplace architecture scales to new cities and content formats by configuration.
- Hours
- AI briefs
- 0-sided
Technologies
React · TypeScript · Supabase · shadcn/ui · Stripe · OpenAI · Tailwind