Bárbara M. Peres · Full-Stack Developer & AI Engineer, Portugal
The interactive version of this page did not load. Reloading usually fixes it. Otherwise the links below still reach me.
Full-Stack AI Nutrition App · 2025
Making healthy eating as simple as taking a photo
A nutrition app with a Flask 3 back end of 28 REST endpoints and a React 18 + Tailwind front end built mobile-first. Azure OpenAI does the model work: GPT-4o reads a food photo and names the dish and its ingredients, an ingredient pipeline handles PT, ES and EN, and DALL·E illustrates the recipes it generates. It also estimates a meal from free text, logs meals and weight, and keeps a gamification layer of XP, levels and badges. It later became the NourishGraph dissertation.
Flask 3 · React 18 · Azure OpenAI · GPT-4o · DALL·E · Tailwind CSS · SQLAlchemy · lucide-react · Gamification
Logging meals by hand is tedious, and most people quit. The brief: tracking as easy as taking a photo, in Portuguese, Spanish or English, with game mechanics to bring users back.
A four-person course project in which I wrote the application, front to back. A Flask 3 REST API of 28 endpoints over Azure OpenAI, with GPT-4o prompts tuned to pull dishes and ingredients out of a single photo, DALL·E for recipe imagery, and an ingredient pipeline that works in PT, ES and EN. The React 18 + Tailwind front end is mobile-first and carries the gamification layer: XP, levels, badges.
The demo works the whole way through: photo in, ingredients detected, calories and macros out, then AI recipes with generated imagery, in three languages. The concept went on to become the NourishGraph dissertation.