GRAiD

Inspiration

As experienced TAs and tutors for our university's Computer Science department, my group and I understand firsthand the challenges of grading projects and providing comprehensive feedback. Spending hours deciphering and debugging walls of code can be not only time-consuming but also mentally taxing. Recognizing the need for a more efficient and accurate grading solution, we developed our platform. GRAID streamlines the grading process, leveraging AI technology to provide precise feedback and suggestions, saving valuable time for both educators and students. With GRAID, we aim to revolutionize the way coding projects are assessed, making the learning experience more effective and rewarding for all involved.

What it does

GRAID is a state-of-the-art platform revolutionizing the grading process for classes with its advanced features. In addition to streamlining the grading process through AI technology, GRAID incorporates a sophisticated AI webcam proctoring feature. This feature ensures exam integrity by monitoring students' activities during exams, providing a secure and reliable environment for assessment. With GRAID, educators can trust that their exams are conducted with integrity while saving valuable time previously spent on manual proctoring. GRAID serves as a robust competitor to platforms like Canvas by offering a comprehensive solution for students and graders alike. Not only does GRAID facilitate the seamless submission and grading of assignments, but it also provides a platform for students and graders to communicate effectively through messages and discussion posts. What sets GRAID apart is its integration of advanced AI-powered features, which streamline the grading process and enhance the overall learning experience. With GRAID, users can expect efficient assignment management coupled with intelligent automation, making it a game-changer in the realm of educational technology.

How we built it

GRAID stands at the forefront of educational technology, boasting a robust foundation developed with the Flutter framework for unparalleled cross-platform accessibility. What sets GRAID apart is its innovative use of OpenAI API to offer advanced AI-powered features, including streamlined grading assistance and personalized feedback. The platform uses a custom model trained using MediaPipe Face Mesh technology, allowing for its AI webcam proctor to accurately flag suspicious activity. This sophisticated technology ensures the highest standards of exam integrity by precisely monitoring students' facial features during coding exams. GRAID's integration of Flutter, OpenAI API, and MediaPipe Face Mesh enhances its capacity to provide a secure, reliable, and user-friendly environment for both educators and students, marking a significant stride forward in the landscape of educational technology.

Challenges we ran into

The platform utilizes a custom model trained using MediaPipe Face Mesh, enabling its AI webcam proctor to accurately flag suspicious activity during exams. However, a notable challenge we encountered during development was effectively handling edge cases in training the AI proctor. Despite this challenge, our team successfully navigated through these complexities to ensure the proctor's reliability and accuracy.

Accomplishments that we're proud of

GRAID is not just a solution to a common problem faced in academic settings; it's a personal triumph for our team. As former TAs and tutors, we understand the frustration of time-consuming grading processes. This personal connection drives our pride in developing a platform that not only streamlines grading and offers accurate feedback but also features a clean and professional UI, despite our initial lack of experience in AI model training. GRAID represents our commitment to solving real-world challenges that impact educators and students alike.

What we learned

Through the development of GRAID, we gained invaluable insights into the complexities of training AI models and the importance of meticulous planning and efficient delegation in managing large-scale projects within tight deadlines. Despite our initial lack of experience in AI model training, we navigated the learning curve with determination, gaining a deeper understanding of the considerations involved in model development. Additionally, GRAID served as a catalyst for honing our abilities to plan and delegate tasks effectively, ensuring the successful execution of a comprehensive solution to a pressing problem in the academic realm.

What's next for GRAID

In our future plans for GRAID, while the platform currently focuses on streamlining the grading process for coding assignments, we envision expanding its capabilities to accommodate a wider range of course materials and assignments. While it theoretically works with any type of assignment, our initial design was tailored specifically with coding assignments in mind. Moving forward, we are committed to enhancing GRAID to cater to diverse academic disciplines and assignment types. Additionally, we aim to enrich the personal feedback provided to students by incorporating custom learning modules and offering unique additional resources tailored to their specific needs, empowering them to excel in their studies and prepare effectively for upcoming exams. In order to reduce the amount of expensive API calls, alternative methods can be used to analyze submissions, including prompt optimizing/engineering.