FitGenius AI

Introduction

In today’s fast-paced world, maintaining personalized fitness and nutrition plans is often complex and fragmented. To solve this, I developed a Personalized Fitness & Nutrition AI Assistant. This tool empowers individuals by analyzing their personal biometrics, lifestyle habits, goals, and specific health notes to generate customized workout routines, nutrition advice, and fitness recommendations, powered by Langflow, Astra DB, and OpenAI models.

Problem Statement

  • Current fitness applications typically offer generalized plans that don’t account for individual nuances like injuries, activity levels, or dietary preferences.
  • Moreover, users often struggle with fragmented platforms — fitness tracking, diet management, and workout planning exist separately.
  • The challenge was to create a centralized, intelligent assistant that adapts to personal profiles and evolving needs in real time.

Objectives

  • Build an AI-powered platform that generates customized fitness advice based on user data.
  • Integrate user personal details, fitness goals, nutritional information, and personal notes (such as injuries, limitations) into the advice generation process.
  • Ensure highly dynamic responses to any user question regarding health, fitness, or nutrition.
  • Seamlessly fetch and process data stored on Astra DB.
  • Provide a scalable, secure, and intelligent platform accessible through a user-friendly UI (built on Streamlit).

Product/Project Analysis

  • Frontend: Developed using Streamlit to enable smooth user interaction — inputs like "profile creation", "add notes", "ask AI", and "view macros".
  • Backend: Profile, notes, and goals are stored and retrieved from Astra DB (serverless cloud NoSQL database).
  • Error Handling: Robust error handling for API timeouts, JSON serializations, and schema mismatches.
  • Data Integrity: Notes with datetime handling are serialized properly to ensure database compatibility.
  • AI Layer:
    1. OpenAI models are fine-tuned with a custom prompt that merges profile, notes, and goals into the query context.
    2. Langflow APIs are used to orchestrate AI logic flows and ensure customized advice.

Architecture

🖥️ Streamlit UI
⬇️
🐍 Python Application Layer
(Main.py, Form Submit, Profiles)
⬇️
🛢️ Astra DB
(Workout Database: Profiles + Notes)
⬇️
🔄 Langflow API
(Flows built using OpenAI Models)
⬇️
🤖 OpenAI GPT Models
  • Notes and Profiles are separately stored in Astra DB for better modularity.
  • Langflow Parser ensures input data is properly stringified before hitting the OpenAI model.

Impact

  • True Personalization: Unlike generic apps, this project dynamically generates advice based on live user data and contextual notes (injuries, dislikes, pain points).
  • Real-Time AI Coaching: The user can ask any question ("What's a safe leg workout if I have knee pain?") and get expert advice based on their unique profile.
  • Scalable and Extensible: Designed in a modular way so it can easily scale to nutrition planning, mental fitness, or lifestyle coaching.
  • Serverless, Cost-Efficient Infrastructure: Leveraging Astra DB ensures zero server maintenance, better uptime, and pay-as-you-go model.

Conclusion

This project bridges the gap between personal trainers, dietitians, and AI by creating a hyper-personalized health assistant. By combining user-specific context (notes + goals + biometrics) and the intelligence of modern AI, it delivers a truly customized, professional-grade coaching experience. It demonstrates expertise in AI system design, cloud infrastructure, database management, and prompt engineering, making it a strong highlight in my professional portfolio.

Let's Connect!

I enjoy connecting with like-minded professionals passionate about technology, strategy, and impact. Feel free to reach out!

Chicago, IL
(312) 871-8022
k.teckchandani1703@gmail.com