Problem
Calorie tracking fails because logging is tedious. People stop weighing food and searching databases within days. The friction, not the willingness, is what breaks the habit.
Approach
CaloriAI replaces manual entry with a photo. A vision language model identifies the dish and estimates the portion, then returns calories and macros with a confidence score, logged against a daily target in one tap. Built EU privacy-first, so meal data stays under European data rules.
Stack
TypeScript · React Native / Expo · vision LLM · FastAPI
Outcome
A working consumer app where a meal goes from camera to logged macros in seconds. The glimpse below shows the capture-to-result flow and the daily nutrition ring.
