The Story
Most nutrition apps lock users into rigid, pre-set meal plans or force them to log food after the fact. EatyWise flips the narrative: it starts with what the user actually craves. The engine scales ingredient portions and adds complementary elements to hit daily macro goals, giving users a dynamic Macro Match Score and 100% freedom to toggle, swap, or tweak ingredients on the fly.
I designed the end-to-end product architecture, mapped the user flows, created the UI design system, and developed the MVP.
A significant focus was backend architecture prior to launch. The goal was to build a pipeline that is reliable, maintainable, and cost-efficient from day one. Every calculation that can be handled deterministically happens in code with verifiable nutritional data, while AI (Google Gemini) runs strictly at the creative culinary layer to pair ingredients and balance macros. Robust fallbacks guarantee predictability if generation misses.
The result is an edge-first architecture designed for real-time interaction — putting user autonomy and desire at the center without sacrificing nutritional precision.
The current version is live in Open Beta for user validation, prioritizing real-usage feedback over assumptions.
- •End-to-end product design, UX architecture, and Next.js/TypeScript frontend.
- •Craving-first Meal Builder with real-time ingredient toggling and Macro Match scoring.
- •Hybrid backend architecture: deterministic macro calculation layer + targeted AI culinary pairing.
- •Built on Next.js, Supabase, and Vercel for high performance and low latency.
- •Fallback-protected pipeline optimized for reliability and cost-efficiency.
Category
Product