Gradiator

• 

  Unpaid Assistants

Team Name: Tech_Bros Members: Mariami Mamageishvili, Anano Tamarashvili, Nini Sharvashidze, Mariami Shonia, Ani Okropiridze.

Inspiration

Grading handwritten assignments is one of the most time-consuming tasks for educators. While accuracy in grading is crucial, providing detailed feedback to every student is equally important—but often impossible due to time constraints. Without proper feedback, students struggle to understand their mistakes and improve. At the same time, professors lack insights into which topics students find most difficult. We wanted to create an AI-powered system that not only automates grading but also analyzes mistakes, provides personalized feedback, and helps educators refine their teaching based on real student performance.

What it does

Gradiator automates the grading of handwritten homework, quizzes, and exams using AI. It recognizes handwriting, evaluates answers, and provides instant feedback to students. Additionally, it analyzes common mistakes and identifies challenging topics, offering valuable insights to professors for improving teaching strategies.

How we built it

We developed Gradiator using machine learning, OCR technology, and NLP to recognize handwriting, evaluate answers, and provide feedback. Our system is trained on diverse handwriting samples to ensure accuracy. The backend processes grading logic, while the frontend provides an intuitive dashboard for professors and students.

Training Process and Localization Our model is trained and tested on previous student works to improve its performance. However, it provides feedback and corrections solely based on the uploaded materials and the specific grading criteria set by the professor. This ensures that grading remains flexible and adaptable to different educational standards while maintaining accuracy and fairness.

Challenges we ran into

Time management and unexpected errors.

What's next for Gradiator

Our next steps include: *Introducing an appeal system where students can request a grade review, allowing graders to reassess their work and make adjustments if needed. *Developing a Georgian handwriting detector to support local language grading. *Expanding functionality to grade more subjects and question types, including code-based assignments. *Integrating with LMS and other educational platforms for seamless adoption. *Enhancing visual analytics to provide deeper insights into student performance and learning patterns.

From Hackathon idea to business :) We strongly believe that our idea is both important and valuable. Therefore, we are committed to continuing its development by improving various features, enhancing the application's convenience, and optimizing the AI model for greater accuracy. Additionally, we see the potential to transform this idea into a business plan with monetization strategies, such as a subscription-based model, offering monthly or annual plans for educational institutions and individual educators. <3

Built With

• api
• fastapi
• gemini
• google-ai-studio
• python
• react
• sqlite
• tailwind
• typescript