Inspiration πŸ’‘

5,000+ test centers in Vietnam. Millions of students grinding 3-hour SATs, IELTS, TOEFL. Every platform measures right or wrong. Nobody asks: was the student even paying attention? We wanted to crack open that black box 🧠

What it does 🎯

FocusIQ reads your brainwaves in real time and adapts your study session.

  • πŸ”” Focus dropping? Instant alert
  • 🧩 Fatigued? AI serves easier questions + hints
  • 😴 Cooked? Smart break based on your actual bio-rhythm

Stick it on. Connect Bluetooth. Study smarter.

How we built it πŸ”§

ECG electrodes from Shopee πŸ›’ β†’ Raspberry Pi Pico running FFT β†’ BLE β†’ Web app that classifies brain power into Focused (>26%), Fatigued (13-25%), or Unfocused (<12%) β†’ OpenAI Codex powers the adaptive AI tutor.

Oh and we flew a 3D printer from Singapore βœˆοΈπŸ‡ΈπŸ‡¬ and borrowed filament from the organizers. Total hardware cost: ~250k VND ($10 USD) πŸ’Έ

Challenges we ran into 😀

The Pico has 264KB of RAM. FFT needs sample buffers. BLE needs its own stack. Running both at once? Constant crashes. We spent hours squeezing buffer sizes, reducing FFT resolution, and staggering memory access so they wouldn't collide.

Codex came in clutch πŸ™Œ helping us debug memory allocation and optimize the FFT to actually fit. Without it we'd still be reading crash logs.

Also: sourcing electronics on Shopee express at midnight in HCMC was a vibe πŸ“¦

Accomplishments that we're proud of πŸ†

  • Full closed-loop neural pipeline built at a hackathon. Real signals, real classification, real adaptation
  • The moment someone picked up their phone and the focus score tanked live on screen ⚑
  • $10 hardware bill
  • Brought a 3D printer on a plane. Yes, really.

What we learned πŸ“š

  • 264KB of RAM teaches you humility real fast
  • BLE timing on the Pico is way more finicky than the docs suggest
  • How Alpha, Beta, Theta waves actually map to cognitive states
  • How to buy ECG electrodes on Shopee at midnight πŸŒ™

What's next for FocusIQ πŸš€

  • Multi-channel EEG headband for better accuracy
  • On-device ML classifier to replace threshold logic
  • Pilot with ETest's 5,000+ students
  • Expand into corporate training and ADHD focus coaching

Built With

Share this project:

Updates