TutorMate

Inspiration

Tutoring involves more than teaching. Student details, lesson notes, assessment results, homework, and practice materials all need attention. TutorMate brings these everyday tasks into one dashboard so tutors can keep records organised and prepare relevant maths practice.

What it does

TutorMate lets tutors create student profiles, manage subjects, record lessons, track progress ratings, and save assessment marks. Lesson records can include homework, notes, points, and payment status.

The worksheet creator supports Primary 1–6 maths, addition and subtraction, three difficulty levels, and 5–20 questions. Tutors can preview worksheets, reveal an answer key, and print the results.

How it is built

The application uses Next.js, React, and TypeScript, with Supabase for authentication and database storage. The web application is hosted on Vercel.

Worksheet generation uses NVIDIA Nemotron through the Nebius Token Factory inference API. The configured model is nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B.

The browser sends worksheet options to an authenticated server route. That route calls the model, checks its response, and returns the questions and answers to the worksheet interface. The Nebius API key stays on the server. Student names, birthdays, and lesson notes are not included in generation prompts.

Work completed during the hackathon

Development during the submission period included the Nemotron-powered worksheet API, configurable worksheet lengths, generation in batches of up to five questions, response validation, retries, duplicate-question handling, printable worksheets and answer keys, and authentication checks for worksheet access.

These additions connect the tutor dashboard to an AI-assisted practice preparation workflow.

Challenges

A usable maths worksheet needs correct answers, the requested number of questions, and distinct equations. A model response alone is not enough.

TutorMate checks JSON structure, question count, allowed arithmetic operations, calculated answers, and duplicate equations. Invalid responses can be retried. Duplicate model output may trigger a deterministic backup generator, and the interface displays a notice when that backup is used.

Accomplishments

TutorMate connects student management, lesson tracking, and worksheet preparation in a working browser workflow. The demonstration shows records being created and saved, followed by worksheet generation, answer-key viewing, and print preview.

The project also keeps model credentials server-side and validates generated arithmetic before presenting it to the tutor.

What the project taught me

Useful AI features need clear boundaries and checks around model output. For this application, authenticated access, arithmetic validation, duplicate detection, and readable printable output matter alongside the generation prompt.

What's next

Next steps include more maths topics, smoother loading states, database setup migrations for easier installation, and clearer progress visualisations.

All student records shown in the demonstration are fictional.

Built With

• nebius-token-factory
• next.js
• nvidia-nemotron
• react
• supabase
• typescript