Inspiration

As educators grapple with evaluating numerous student submissions, we envisioned a solution that automates the grading process while maintaining fairness and accuracy. Inspired by the growing potential of AI in education, we set out to create grAIder—an AI-powered code grader that streamlines evaluation, reduces manual effort, and provides consistent feedback.

What it does

Analyzes and evaluates code accuracy, efficiency, and style. Supports multiple programming languages. Provides detailed annotations and feedback. Flags errors, highlights improvement areas, and assigns grades dynamically.

How we built it

Backend: Flask framework with Python to handle API requests and manage grading logic. AI Model: Leveraged GeminiAPIs to assess code quality and provide contextual feedback. Database: SQLLite to store student submissions, grading rubrics, and evaluation results. Frontend: Tailwind CSS for responsive and intuitive UI/UX design.

Challenges we ran into

Ensuring language-agnostic grading across multiple programming languages. Implementing rubric-based evaluations that adapt dynamically to different assignment criteria. Handling edge cases where code structure or logic was valid but produced unexpected results. Debugging integration issues between Flask, SQLlite, and the AI model to ensure seamless functionality.

Accomplishments that we're proud of

Successfully developed a robust and scalable code grading platform. Achieved accurate evaluation across diverse programming languages. Ensuring language-agnostic grading across multiple programming languages. Implementing rubric-based evaluations that adapt dynamically to different assignment criteria. Handling edge cases where code structure or logic was valid but produced unexpected results. Debugging integration issues between Flask, SQLlite, and the AI model to ensure seamless functionality. Built a clean UI for professors to review results easily.

What we learned

Deepened our knowledge of integrating AI with backend APIs. Mastered Flask and SQLlite interactions for data handling. Gained insights into creating adaptive rubrics and grading mechanisms.

What's next for grAIder

Language Agnostic: The AI decides the language by assignment description. Plagiarism Detection: Implement AI-powered similarity checks. Enhanced Feedback: Improve the depth of feedback with code complexity analysis. Gamified Learning: Introduce badges and leaderboards to encourage coding improvements. Professor Analytics: Provide insights on class performance trends and individual student progress.

https://github.com/EyadAbouKer/grAIder

Built With

Share this project:

Updates