FitVis

From the first idea to running the application

FitVis is my nutrition application for keeping track of meals, nutrients, recipes, weight and hydration. I built the entire project myself from scratch: the user application, backend, data processing, nutrition-adviser portal and internal administration.

Three platforms: web, iOS and Android

I released the application on three platforms: the web, iOS through the App Store and Android through Google Play. My work covered the complete path from building the application to releasing it for users.

The technology behind it

The mobile and web application uses Flutter. Its core is .NET 10 and C#, with separate layers and CQRS through MediatR. Nutrient estimation and food recognition from photos use Python, FastAPI and FAISS. The adviser portal runs on Next.js, while internal administration uses Razor Pages.

I separate data by purpose: MongoDB for the food catalogue and nutrition records, PostgreSQL for the relational core, Elasticsearch for search and autocomplete, and Redis for caching and rate limiting.

900,000 foods and room to grow

On 22 October 2026, the MongoDB catalogue contained 900,000 food items. I designed the architecture with future sharding and larger data volumes in mind.

Deployment is part of my work too

I run the individual components on my own VPS using Docker Compose and automate deployment with GitHub Actions. The project covers application development, architecture, a substantial database and its operational setup.