AmpliTutor

Inspiration

Identity isn't something you find—it's something you build. And learning is how you build it.

Growing up, we’re put into labels. “The math kid." "The B-student.” And here's the problem, we internalize those labels and they become how we see ourselves. They become the ceiling on who we can become.

Our team noticed this firsthand; despite using the same study tools and techniques, different people excelled in different ways.

This is where our idea came from. We wanted to build a learning product that recognizes identity as something expressed through behaviour, protected through adaptation, created through feedback, and evolving over time. Instead of forcing users to adapt to the system, we will make the system adapt to them, protecting their confidence, reinforcing their strengths, and evolving alongside them as their learning habits change.

What it does

As you grow, your tools should grow with you. AmpliTutor is a dynamically evolving tool that self-improves based on your behaviour. Every session you enter helps AmpliTutor and helps the future you. AmpliTutor tracks your time, accuracy, and the subtle nuances of your performance and behaviour through visual cues and your interactions with the computer, mapping your unique learning DNA. It absorbs these data points in real-time, constantly recalibrating your personalized data vectors to evolve its understanding of how you learn. By transforming your history into a blueprint for mastery, the system engineers a tailored problem set and analytics designed specifically for your "sweet spot." AmpliTutor decodes and displays your habits so you can learn faster, retain more, and dominate your goals with scientific precision.

How we built it

AmpliTutor is powered by: Frontend and UI:

Next.js, HTML, Tailwind CSS AI & Machine Learning:
PyTorch for user-specific type-responsiveness tuning
OpenCV for detecting user focus and attention
Google Gemini (via OpenRouter) for question generation and validation Backend and Data:
Python
FastAPI
Postgres

Challenges we ran into

We were originally building the frontend and UI with React, but we realized that it would be difficult to connect it to the backend, so we had to redo everything with Next.js
In a bit of a funny but tragic moment, we spent SO LONG trying to figure out why our camera wasn’t working -> turns out it was covered THE WHOLE TIME

Accomplishments that we're proud of

Learning to use many new tools and frameworks like PyTorch and OpenCV to create something genuinely useful and impactful
Not rushing into building but instead planning out the system was an amazing choice which saved us a lot of headache and wasted work. Taking the time to make sure we understood what and why we were doing and proactively searching for mistakes prevented us from pursuing lost causes.
Creating a functional, polished tool that we can use ourselves (as we've done many time for some reason during development)

What we learned

Fully flush out each others ideas before the building phase to identify any weaknesses or flaws - not as attacks on each other, but to make your approach bulletproof