CoursePilot AI

Inspiration

University course selection can be confusing and stressful. Important information is often scattered across training plans, course-selection guides, academic records, and semester notices. Students may know how many credits they have earned, but still struggle to understand which requirements remain incomplete and which courses should be prioritized next.

I started CoursePilot AI during this hackathon to build a clearer and more accessible personal course-planning assistant for university students.

What it does

CoursePilot AI helps students understand their academic progress and plan future course selections.

Students can upload their training plan, completed courses, and current course selections. The system compares this information with graduation requirements, elective modules, general-education categories, practical courses, and optional postgraduate-planning requirements.

It separates courses into completed, currently selected, lottery-pending, and currently available categories. The platform then shows completed requirements, missing courses or credits, suggested priorities, possible scheduling conflicts, and items that still require manual confirmation.

Each recommendation includes an explanation so students can understand why a course is considered necessary or important.

How we built it

We designed CoursePilot AI as a full-stack web application using Next.js, TypeScript, React, PostgreSQL, Docker, and GitHub.

Training plans and course-selection documents are converted into structured course and rule data. The analysis system matches student course records with these rules and calculates progress across different academic modules.

Personal academic data is intended to remain on the student's device. The server stores only public rules and approved training-plan information, while uploaded files are used temporarily for analysis.

Challenges we ran into

The main challenge was translating complicated academic documents into rules that a computer could evaluate reliably.

Requirements may change according to a student's major, admission year, course category, training-plan version, and personal goals. Some requirements are mandatory, while others are only recommendations. Course availability, capacity, scheduling conflicts, and lottery results may also change each semester.

Another challenge was handling uncertain or incomplete information without presenting misleading conclusions. To reduce errors, the system separates confirmed results from uncertain information, displays the evidence behind recommendations, and asks students to review automatically recognized course data.

Accomplishments that we're proud of

We are proud that CoursePilot AI was started from scratch during this hackathon and developed from a real student problem into a structured software project.

We created a clear product scope, designed the full-stack architecture, organized the main data sources, and defined how training plans, completed courses, current selections, and course-selection rules should work together.

We also designed the system around transparency and privacy. It does not require students to provide their academic-system passwords, and it avoids presenting uncertain analysis as an official academic decision.

What we learned

Through this project, we learned how frontend pages, backend services, databases, file parsing, and rule-based analysis work together in a full-stack application.

We also learned that a useful AI-assisted education tool should not simply generate answers. It should explain why a recommendation was made, identify the source of each conclusion, and clearly show which information still requires manual confirmation.

The hackathon helped us learn how to reduce a large idea into a realistic first version and build around the needs of actual users.

What's next for CoursePilot AI

Next, we plan to complete the course and rule database, improve document and screenshot parsing, and build the main student analysis interface.

We also plan to add course-search and filtering tools, timetable-conflict detection, clearer progress visualizations, and more detailed explanations for course recommendations.

In the longer term, CoursePilot AI could support more training-plan versions, majors, and institutions while continuing to protect student privacy and keep every recommendation transparent and verifiable.

Built With

• ai
• codex
• docker
• github
• next.js
• openai
• postgresql
• react
• typescript