Teen Calculus AI Trainer

I am building an AI-powered interactive tutor that helps 13- to 15-year-olds strengthen their calculus problem-solving skills. The system should generate step-by-step practice questions, evaluate student responses in real time, and offer targeted hints rather than full solutions to keep learners actively engaged.

Core scope
• Curriculum focus: limits, derivatives, and basic applications of calculus appropriate to early high-school level.
• Learning mode: dynamic problem sets that adapt in difficulty based on each student’s recent performance, tracking accuracy and time spent per step.
• Feedback engine: concise, conversational explanations that highlight common misconceptions without overwhelming the student with theory.
• Interface: web-based (React or similar) or mobile-friendly, with a clean dashboard for students to review progress and for teachers/parents to download simple CSV reports.

Key functions to implement
1. Problem generator that tags each question by topic and difficulty.
2. Automatic grading with partial-credit logic for multistep questions.
3. Hint algorithm that surfaces one hint at a time, increasingly specific.
4. Progress model that recalibrates the next set after every completed quiz.
5. Admin panel to add new questions and edit existing ones without code changes.

Acceptance criteria
• At least 200 unique calculus practice items seeded in the database.
• Accuracy of automated grading ≥ 95 % across a supplied test set.
• Response time under 1 s for generating and grading a typical problem.
• Deployed demo hosted on a free tier (Heroku, Vercel, or similar) for review, with source code pushed to a Git repository.

If you have prior experience with adaptive learning engines, NLP-based feedback, or educational front-end design, that background will be invaluable. The final deliverable is the fully functional platform plus installation and usage documentation that a non-technical teacher can follow.