repD

A full-stack fitness tracking platform that integrates with Telegram to log workouts seamlessly. Uses a custom NLP parsing engine backed by Gemini AI to interpret natural language workout messages and visualizes progress on an interactive web dashboard.

Timeline

May 2026 - Present

Role

Full Stack Developer

Team

Solo

Status

Completed

Live Demo Source Code

repD : AI-Powered Fitness & Workout Tracker via Telegram

What Users Can Do

Telegram Bot Integration: Log workouts instantly through a Telegram bot using natural language (e.g., 'bench press 3x10 60kg').
Smart NLP Parser: Accurately interprets various workout string formats and extracts exercise, sets, reps, and weight.
AI-Powered Analysis: Integrates Gemini API to analyze complex or malformed inputs for resilient workout tracking.
Progress Visualization: An interactive web dashboard built with React and Recharts to visualize workout progress over time.

Tech Stack

Reactjs
Vite
Tailwind
Express
PostgreSQL
Gemini
NodeJs

Challenges I Faced

Engineered a robust regex-based NLP parser with Gemini fallback to process unstructured natural language inputs from Telegram.
Managed state sync and secure authentication between the web dashboard and the Telegram webhook system.

How I Overcame Them

Integrated serverless databases (Neon) and modern ORMs (Drizzle) for efficient data modeling.
Built resilient webhook endpoints to handle asynchronous messages from Telegram reliably.