Cougar Degree

An AI-assisted degree-planning prototype and CodeRED Astra winner for Best Use of DigitalOcean Gradient AI.

Comment 2

Cougar Degree

Cougar Degree is an AI-assisted degree-planning prototype built by Mahmoud Badi, Michael Monroe, and Rami Abdelrazzaq at CodeRED Astra. The project won [MLH] Best Use of DigitalOcean Gradient AI.

Inspiration

Planning a degree means coordinating requirements, prerequisites, completed courses, and semester workload. Inspired by CougarGrades, our team explored a simpler interface for turning that information into a semester-by-semester starting plan for University of Houston students.

What it does

The interface accepts a major, completed coursework, and planning preferences, then presents a proposed degree path with course and credit breakdowns. The project explores eight-semester planning and credit-load balancing around 15 hours per semester.

How we built it

The application uses a vanilla JavaScript/HTML/CSS frontend and a Node.js/Express backend. The backend includes API routes for plan generation and major listings, structured prompts, response parsing, rate limiting, Helmet, and CORS configuration.

The hackathon-winning version integrates an AI agent through DigitalOcean Gradient AI. Its repository README documents the llama3-8b-instruct model and grade-aware planning goals. The original Devpost story describes a Gemini-based implementation; the project update and linked winning-version branch document the DigitalOcean version. These are distinct versions of the team's project.

Team and my role

I was a member of the three-person hackathon team. This page presents our collaborative project and its award; implementation credit remains shared with Mahmoud Badi and Michael Monroe.

Challenges and learning

The project brought together structured course information, AI output, and a usable planning interface. The key challenges were coordinating frontend/backend behavior, prompting for consistent plan structures, and making prerequisites and semester loads understandable to students. We learned how much the usefulness of an AI planner depends on the quality of its underlying course data.

Accomplishment

Winner of [MLH] Best Use of DigitalOcean Gradient AI at CodeRED Astra, as recorded on the project's Devpost page. The public repository preserves the winning version for review.

Project status and next steps

Hackathon prototype. Generated plans are starting points for review with an academic advisor; they do not guarantee graduation or replace official degree requirements. Future work includes stronger validation against current catalogs, clearer handling of substitutions and transfer credits, and saved plans.

Explore

Created by Mahmoud Badi, Michael Monroe, and Rami Abdelrazzaq.

Share this project:

Updates

posted an update —

Digital Ocean Gradient AI Agent Version: michael-3 Branch

What Makes This Branch Special

The michael-3 branch introduces an AI-powered degree planning system that balances course difficulty across semesters using real grade distribution data.

AI Agent on Digital Ocean's Gradient AI Platform

  • Deployed using DigitalOcean's Gradient AI infrastructure with the llama3-8b-instruct model
  • RESTful API integration with Bearer token authentication
  • Automated agent discovery and configuration scripts

CougarGrades.io Integration for Course Difficulty The system leverages cougargrades.io's public grade distribution data to:

  • Calculate average GPA per course
  • Classify course difficulty levels based on historical performance
  • Make data-driven decisions about course combinations

Smart Semester-by-Semester Difficulty Balancing The implementation uses a sophisticated balancing algorithm that:

  • Targets 15 credits per semester using bin-packing logic
  • Automatically resolves prerequisite dependency chains before course placement
  • Evaluates each semester's overall difficulty and redistributes courses to prevent overwhelming workloads
  • Generates complete 8-semester degree paths from start to graduation

Technical Architecture

  • Node.js/Express backend with rate limiting and security middleware
  • 130+ line structured AI prompt engineering that guides the model to consider workload balance, prerequisites, and course categories
  • Vanilla JavaScript frontend with real-time plan visualization
  • RESTful API design with health checks, major listings, and plan generation

The Innovation Unlike traditional course planners that only check prerequisites, this AI agent considers:

  1. Historical course difficulty from real grade distributions
  2. Cognitive load balancing across semesters
  3. Prerequisites and course sequencing
  4. Credit hour optimization

The result: Students get personalized degree plans that are not just feasible, but manageable, preventing burnout semesters while maintaining steady graduation progress.

Log in or sign up for Devpost to join the conversation.

Submission history