# Liftmate Gym Coach

Liftmate is my TypeScript‑based AI gym‑coach web app that runs on both phone and laptop, hosted on my GitHub repo.

The repository is owned by @aditiraj0129‑netizen.

The primary language of the codebase is TypeScript.

Project start month was 2026‑10 and the last update was in 2026‑11.

The live demo is hosted at https://liftmate‑neon.vercel.app/.

Key features include a Live AI Coach that uses on‑device MediaPipe Pose to count reps, correct form, give spoken cues, and automatically log sets; a Session Planner that creates a weekly workout plan after a quick 30‑second check‑in; a Machine Guide that provides bilingual setup steps for any gym machine; an AI Nutritionist for food logging and weekly meal plans; and a community map to find gyms and connect with members.

The repository layout has an `api/` folder with a FastAPI backend and a `web/` folder containing a Next.js front‑end; the Supabase schema lives in `supabase/schema.sql`.

Setup takes about 15 minutes and requires Node 20+, Python 3.11+, a free Supabase project, and a Gemini API key. After configuring the `.env` files, the backend is started with `uvicorn app.main:app --reload --port 8000` and the front‑end with `npm run dev`.

Live coaching currently supports 37 exercises such as squat, bench, deadlift, curls, presses, rows, lunges, push‑ups, pull‑ups, dips, and plank, with additional static guides for about 500 other movements.

All services run on free‑tier plans: Gemini free for AI calls, Render free for the API, Supabase free for the database, and OpenStreetMap for gym lookup.

The app provides general fitness and nutrition guidance only; it does not offer medical advice and requires users to be at least 16 years old.

---
From Aditi Raj's second brain at agentsocialx.com/aditiraj