KatChef

Auth Screen
Home Screen (1)
Home Screen (2)
Home Screen (3)
Scan Screen
Profile Screen (1)
Profile Screen (2)
User Preferences

Inspiration

A lot of people already have food at home, but still end up asking the same question: “What can I actually cook with this?” Too often, ingredients get forgotten in the fridge, people buy duplicates, and perfectly usable food gets wasted simply because there is no clear plan.

We wanted to build something that solves both problems at once: reduce food waste and reduce decision fatigue. That became KatChef, an AI-powered kitchen companion that helps people scan ingredients, organize a virtual fridge, and turn what they already have into realistic meals.

What it does

KatChef is an AI-powered cooking companion with three core flows:

KatLens scans a food photo and detects likely ingredients.
MyFridge stores and organizes ingredients in a searchable virtual fridge.
Chat with KatChef turns those ingredients into realistic meal ideas, recipe suggestions, and practical cooking advice.

The goal is simple: help people cook more from what they already have, waste less food, and make everyday meal decisions feel easier.

We wanted KatChef to feel like something people would actually want to use, not just a quick AI prototype, so we put a lot of care into the UI, responsiveness, and overall experience.

How we built it

We built KatChef as a full-stack cross-platform product:

Frontend: Expo, React Native, React Native Web, TypeScript, Zustand, React Navigation
Backend: FastAPI, Python
Auth + database: Firebase Auth and Firestore
AI: Gemini for meal guidance and recipe generation
Vision: Google Cloud Vision for ingredient detection

On the frontend, we built a shared design system with a warm cream/coral/green palette, responsive layouts, mobile-friendly touch targets, and platform-aware flows for web vs mobile.

On the backend, we created API routes for chatbot and scan detection, added health diagnostics, deploy-safe configuration, CORS handling, timeout handling, and fallback logic so the product stays resilient even when external AI services are slow or imperfect.

Challenges we ran into

One of the biggest challenges was making the same product feel good across multiple environments:

web browser
Expo mobile app
native-style mobile flows
deployable backend APIs

We also ran into real-world integration issues:

Firebase auth behavior across web and native
Google Vision billing / API setup
Gemini response speed and malformed JSON edge cases
balancing rich AI output with fast, product-quality response times
making scan and chat flows fail gracefully instead of feeling broken

A lot of the work ended up being about turning a cool prototype into something that actually behaves like a product.

Accomplishments that we're proud of

We’re proud that KatChef is more than a concept demo. It became a real cross-platform product with:

a polished web experience
a mobile-ready Expo codebase
a deployable FastAPI backend
real Firebase user persistence
AI chat that generates practical recipe ideas
ingredient scan flow with browser-safe fallback behavior
profile, badges, preferences, and a more complete product loop

Most importantly, it tackles a real everyday problem: helping people use what they already have before it turns into food waste.

What we learned

We learned that building a strong AI product is not just about calling a model. The hard part is product engineering:

designing reliable flows around imperfect AI output
handling auth and platform differences cleanly
making failures understandable to users
tuning latency so the experience still feels fast
keeping the interface polished across web and mobile

We also learned how much polish matters. Small UI, state, and error-handling decisions make a huge difference in whether an app feels like a hackathon prototype or a real product.