📖 About the Project: BodyTrust AI
🔥 What Inspired Me
The idea for BodyTrust AI came from a simple question:
"What if we could earn trust in our physical health the same way we build financial credit — with data?"
As someone who understands the growing connection between wellness, technology, and analytics, I wanted to create a system that didn't just track steps and calories, but turned raw activity into intelligent, actionable health insights.
This project was born during the PhysTech 2025 Hackathon, but quickly became something much bigger — a personal mission to blend everything I’ve learned into one meaningful build.
🧠 What I Learned
This was my most challenging and rewarding solo project so far. I learned:
- How to design and integrate a full-stack application using FastAPI + Dash
- How to apply machine learning to real-world health data
- How to handle authentication, session state, and role-based access in production-ready systems
- The importance of data cleaning and normalization for accurate predictions
- How to structure code and documentation like a professional open-source project
It also taught me that confidence comes from shipping, not certificates.
🛠️ How I Built It
- Backend: FastAPI with secure JWT login and prediction routes
- ML Model: RandomForestRegressor trained on Fitbit activity + sleep data
- Frontend: Plotly Dash for public and personalized dashboards
- Database: PostgreSQL with SQLAlchemy ORM for user and role handling
- Admin Panel: Role-based access with tools to promote users
- Score Engine: Weighted formula + clustering using KMeans
- Security: Environment variables, password hashing, and git hygiene
- Docs: Professional README, live dashboards, and a full system PDF
⚔️ Challenges I Faced
- 💾 Merging complex multi-source data into a single usable format
- ⚖️ Finding the right features for both clustering and prediction
- 🔄 Syncing real-time user sessions between Dash and FastAPI
- 🔐 Implementing secure JWT auth and admin-only visibility
- 🔗 Keeping everything modular, reusable, and production-grade
These weren’t just bugs — they were opportunities to level up.
🎯 Final Words
BodyTrust AI isn't just a hackathon project — it’s a working vision of smarter health tracking.
I didn’t build it to win a prize — I built it to prove to myself that I could bring together everything I’ve learned and solve a real-world problem.
And I did.
With data.
With code.
With fire. 🔥
Built With
- dotenv
- fastapi
- joblib
- jwt
- kmeans
- numpy
- oauth2passwordbearer
- pandas
- passlib
- plotly-dash
- postgresql
- python
- randomforestregressor
- scikit-learn
- sqlalchemy
Log in or sign up for Devpost to join the conversation.