F/AI — AI fitness trainer, 340,000+ users by Sevastian RakhimovF/AI — AI fitness trainer, 340,000+ users by Sevastian Rakhimov

F/AI — AI fitness trainer, 340,000+ users

Sevastian Rakhimov

Sevastian Rakhimov

What it is

F/AI is an AI fitness trainer: it builds a personalised training plan, adapts it as you progress, reads your body from a single photo, and answers questions in chat like a coach would. It ships on iOS and Android and it is a paid product, not a demo.
I am the co-founder and CTO. I wrote the first production version of the Flutter app alone over the first five months, and I still own the whole stack end to end — mobile, backend, infrastructure, releases and the pager.
F/AI workouts screen
F/AI workouts screen

Where it got to

340,000+ registered users — the counter on fitgpt.pro/stats is live, so you can check it yourself instead of taking my word for it
23,000 monthly active users
$8,000 MRR across the App Store, Google Play and direct web subscriptions
2,000+ active paid subscriptions
4.5★ on the App Store

How it is built

Mobile. Flutter for the app, with native Swift dropped in where the camera and health integrations demanded it. Apple Health on the input side, so recovery and load are read from real activity rather than self-reporting.
Backend. Supabase and PostgreSQL underneath, Python and FastAPI for the services that do the heavy work, Docker on AWS, CI/CD, monitoring and alerting.
The AI part. LLM pipelines on LangGraph rather than a prompt in a loop: plan generation and editing over the user's own history, a computer-vision body scan, and a chat coach that can look at a photo of a gym machine and tell you what to do with it. Evaluation and observability wired in from the start, because an AI feature that nobody measures is an AI feature that quietly degrades.
Money. Subscriptions across three billing sources — App Store, Google Play and a separate web checkout — that have to reconcile against each other, with RevenueCat holding entitlements in one place.
Chat with your AI Coach
Chat with your AI Coach

Why this is the case study I lead with

Most AI portfolios show a prototype. This one has been in the stores for two years with real users who complain when something breaks, real money moving through three payment sources, and a support inbox. The interesting engineering was never the model call — it was auth, billing reconciliation, error handling, cost control and observability. That is the half I get hired for.
The app is available in English and Russian: App Store · Google Play
Like this project

Posted Aug 24, 2026

Co-founder and CTO. Built the app, the backend and the LLM pipeline, and took it from zero to 340,000+ users, 23,000 monthly actives and $8,000 MRR — live in both stores.