Case Study · Shipped Product
Gud For Us — AI Ingredient Scanner
AI-powered mobile app for analyzing food and cosmetic products, delivering ingredient insights, health scoring, and personalized compatibility.
Helps users make better product choices by turning complex ingredient data into simple, personalized insights.
Tech Stack
React NativeExpoTypeScriptNativeWindSupabaseReact QueryGemini AI
Role
Frontend Owner · Product Contributor
Core Contribution
- 01Built the entire frontend using React Native (Expo + NativeWind) and TypeScript — from blank canvas to shipped app.
- 02Designed and implemented onboarding, product scanning flow, and results UI with a focus on clarity and speed.
- 03Implemented a personalized user flow that adjusts AI-generated results based on inputs collected during onboarding.
Engineering Decisions
- 01Adopted React Query for all API state — eliminated redundant network calls and gave the UI optimistic, cache-aware behavior.
- 02Designed product deduplication using unique slugs, preventing repeated scan entries and keeping the data layer clean.
- 03Structured Gemini AI responses into a consistent format before rendering, decoupling UI components from raw AI output variability.
Product Thinking
- 01Built a compatibility scoring system tied to the user's onboarding profile — same product, different result per user.
- 02Collaborated with backend to establish a product → article pipeline, enabling SEO-indexed content pages from scanned data.
- 03Simplified ingredient complexity into three clear UI categories (beneficial / neutral / concerning) to reduce cognitive load.
Challenges
- 01Ingredient data across product formats is highly inconsistent — solved by normalizing AI output through a strict response schema before it reaches the UI.
- 02AI responses can vary between calls for the same product — handled by caching structured results and flagging low-confidence outputs for review.