Skip to content
Full-StackMobile App

kerak

AI-powered personal nutrition assistant. Calorie tracking, recipe generation, weekly meal planning and nutrition advice - all driven by Mistral AI.

Demo

The Problem

Tracking nutrition manually is tedious - counting calories, balancing macros, finding recipes that fit your goals. Existing apps are either too complex, filled with ads, or lack intelligent suggestions. There's no simple tool that understands what you eat in plain language and builds a personalized plan around it.

The Approach

Kerak lets you log meals in natural language - just say what you ate and the AI figures out the rest. It generates recipes from your ingredients, builds weekly meal plans based on your goals, and gives personalized nutrition advice. The whole experience is conversational and effortless.

Technical Choices

Mistral AI for everything

Natural language meal parsing, recipe generation from ingredients, weekly routine planning, and nutrition analysis - all powered by Mistral Small. Structured JSON output keeps the responses reliable and parseable.

React Native + Expo

Cross-platform mobile app with NativeWind for styling. Zustand for state management keeps things simple and performant. Expo handles the build pipeline and push notifications.

Express + Prisma + PostgreSQL

Clean REST API with Prisma ORM for type-safe database access. CIQUAL ingredient database (~10,000 entries) for accurate nutritional data. Node-cron generates fresh daily recipes every morning at 5 AM.

Key Features

Natural language meal logging with AI parsing
Recipe generation from your ingredients
Weekly meal plans based on your goals & budget
Macro tracking with personalized nutrition advice
Smart suggestions from favorite recipes
Step-by-step cooking mode with progress
Daily auto-generated recipes (5 AM cron)
Customizable meal reminders & themes

Tech Stack

Mobile

React Native 0.81, Expo 54, Zustand, NativeWind, react-native-chart-kit

Backend

Express 5.1, Prisma 7.6, node-cron, JWT auth

AI

Mistral Small (recipe gen, meal parsing, nutrition advice)

Database

PostgreSQL 16, CIQUAL ingredient database (~10,000 entries)

SourceGitHub