StudyShield

StudyShield predicts your burnout risk using AI and builds a personalized, burnout-safe daily schedule — so students can work hard without burning out.

Comment

Inspiration

I'm a high school student juggling 4 AP classes, 2 dual credit courses, a part-time job, volunteering, and independent AI research — all while working toward my goal of becoming an AI engineer. I've felt the weight of burnout firsthand, and I noticed there was no tool built specifically for students like me. So I built one myself.

What it does

Burnout Shield is a two-part student wellness app:

  1. Burnout Risk Predictor — a neural network trained on student habit data (sleep, workload, screen time, mood, caffeine, and physical activity) that predicts your burnout risk level (Low / Medium / High) and delivers personalized, AI-generated tips to help you course-correct.

  2. AI Study Scheduler — an intelligent daily planner where you input your school hours and to-do list with due dates and estimated times. The AI generates a structured, burnout-safe after-school schedule with built-in breaks, prioritized tasks, and a wind-down routine — because rest is part of the plan, not an afterthought.

How we built it

  • ML model — built and trained in PyTorch with a fully connected neural network on synthetically generated student lifestyle data
  • LLM integration — used the Groq API (Llama 3.3 70B) as a free, fast inference layer for personalized tip generation and schedule creation
  • Frontend — clean, responsive UI built in HTML/CSS served through a Flask backend

Challenges we ran into

  • Finding a truly free LLM — most APIs have paywalls or strict rate limits; Groq's free tier with Llama 3.3 70B was the breakthrough
  • Prompt engineering the scheduler — getting the AI to produce a structured, burnout-conscious schedule (not just a task list) required careful prompt design around break rules, priority ordering, and wind-down time
  • Bridging Python and the browser — wiring a PyTorch model and LLM API calls into a seamless Flask + HTML/CSS frontend was a full-stack challenge I hadn't tackled before

Accomplishments that we're proud of

This is a tool I genuinely plan to use every day. Building something that solves a real problem I live with — and that could help other overloaded students — makes this more than a hackathon project. I'm proud that it works end-to-end: real ML, real AI, real UI.

What we learned

  1. How to train and deploy a neural network with PyTorch inside a web app
  2. How to integrate and prompt-engineer a large language model via API
  3. How to build a full-stack app connecting Python logic to a polished HTML/CSS frontend
  4. That shipping something useful under pressure is itself a lesson in avoiding burnout

What's next for StudyShield

Next, I plan to add calendar sync so students can import assignments directly, a mobile app for on-the-go access, study streak tracking to build consistency, and a dashboard that monitors burnout trends over time so students can spot warning signs before they hit a wall.

Built With

Share this project:

Updates