Coach App — AI platform for personal fitness support

Stack:
Python, FastAPI, Django, PostgreSQL, Redis, Celery, WebSockets, Docker, LangChain, LangGraph, Qdrant, JWT, REST API

Project Description:
Development of an AI platform for personal user support. The system combines a classic backend for storing profiles, workouts, progress, and plans with an AI module that analyzes user data and generates personalized recommendations. The platform was designed as a scalable product with a microservices architecture, where individual services are responsible for authorization, fitness logic, AI interaction, and real-time communication. The main task is to provide the user not just an activity tracker, but a full-fledged AI coach that remembers history, considers goals, physical parameters, and progress dynamics.

What has been implemented:

user registration, authorization, and profile system
storage of anthropometric data, goals, and user parameters
training module, exercises, plans, and activity history
AI chat for interaction with a personal fitness coach
storage of dialogue memory and recommendation history
generation of personalized workout and load recommendations
asynchronous task system through Celery
JWT authorization and API for client applications
containerization of all services through Docker

My responsibilities:

designing the architecture of the entire system
developing the backend part and API
building the microservices structure
implementing the logic of the AI coach
integrating LangChain / LangGraph / Qdrant
implementing WebSocket communication for chat
designing PostgreSQL schemas and relationships between entities
connecting Redis and Celery for background processing
Docker-based deployment and infrastructure preparation for scaling