Paper App – AI-Powered Assessment Design System

Paper App is an AI-powered assessment design system that uses Gemini 3 to create fair, outcome-aligned exam papers from PDFs, notes, or topics — with controlled

Comment

Inspiration

Designing fair and outcome-aligned examination papers is one of the most challenging tasks in education. Educators must balance CLOs and PLOs, difficulty levels, marks distribution, and question types, yet most existing tools only focus on generating questions. I wanted to solve this deeper problem by treating exam creation as an assessment design challenge, not just a content generation task.

What it does

Paper App – AI-Powered Assessment Design System creates complete, structured exam papers from syllabus PDFs, pasted notes, or simple topic inputs. It supports both outcome-based (CLO/PLO-aligned) and simple modes, allows control over difficulty levels (easy, medium, hard), flexible marks distribution, scenario- based questions, and customizable paper headers (semester, department, date, etc.). The final output can be copied or exported as a DOC file, ready for real academic use.

How we built it

The project was first prototyped using Gemini 3 through its canvas environment and gradually refined through iterative prompt engineering. Gemini is used as the core reasoning engine to analyze long-context inputs such as PDFs and course notes, and to design outcome-aligned assessments under defined constraints.

Once the logic was validated, Gemini was also used as a development assistant to help refactor the code into a clean project structure. The application was implemented using Node.js and React, packaged as a desktop app with Electron, and compiled into an executable installer using Inno Setup for smooth local execution. Gemini was also used as a creative assistant to generate visual assets.

Challenges we ran into

One major challenge was maintaining consistent AI behavior. At times, Gemini produced incorrect or unbalanced outputs, which required repeated prompting, stricter constraints, and multiple iterations. Another challenge was keeping the user interface simple while enforcing complex assessment logic in the background.

Accomplishments that we're proud of

I am proud of building a system that goes beyond question generation and genuinely supports outcome-based assessment design. Paper App produces exams that are structured, aligned, and ready for real academic environments, saving time while improving assessment quality.

What we learned

This project reinforced the importance of treating AI as a reasoning partner rather than a replacement. Clear prompt design, validation, and human oversight are essential for building reliable educational AI systems. Long-context reasoning proved especially powerful for syllabus-level understanding.

What's next for Paper App – AI-Powered Assessment Design System

Future plans include LMS integration, advanced analytics for CLO coverage and exam quality, collaborative exam design for faculty teams, and enhanced explainability to further strengthen trust in AI-assisted assessment design.

Built With

Share this project:

Updates