TutorMind_AI

Inspiration

As an engineering student prepping for placements, I felt the gap between generic courses and real personalized teaching — so I built a tutor that actually adapts to you.

What it does

Diagnoses your skill level, teaches personalized lessons, grades your homework like a real teacher, and autonomously decides your next step — repeat, advance, or review.

How we built it

Node.js + Express backend, Gemini API for tutoring intelligence, Google Cloud Firestore for student data, KaTeX for math rendering, Razorpay for payments, deployed on Render.

Challenges we ran into

Free-tier Gemini quota limits during testing, JSON responses getting truncated mid-generation, and picking a GCP setup that stayed genuinely free while still using real cloud infra.

Accomplishments that we're proud of

Getting the AI to make real autonomous pedagogical decisions — not fixed rules, but Gemini reasoning through a student's history to decide what happens next, live in production.

What we learned

Building an "AI-native" product means designing for AI to make decisions, not just generate content — and that structured, autonomous reasoning is harder than it looks.

What's next for TutorMind_AI

Real payment verification via webhooks, a structured curriculum tree instead of free-text topics, multi-turn chat for follow-up questions, and onboarding real paying students.

Built With

• express.js
• google-cloud-firestore
• google-gemini-api
• html/css
• javascript
• katex
• node.js
• razorpay
• render