TigerTrack — Alyan Anwar

Overview

TigerTrack is an AI-powered academic planning workspace built for University of the Pacific students. We built it at PacificHacks — UOP's annual hackathon — in 11 hours as a 3-person team.

The problem it solves: UOP students have to jump between DegreeWorks (to see what they've completed), College Scheduler (to search for courses), and their own notes just to answer basic questions like "what can I take next semester?" TigerTrack puts all of that into one place — with an AI layer on top.

The app shipped with 988 real UOP catalog entries across 77 subjects, live plan persistence per user, a prerequisite-aware algorithmic scheduler, and an AI chatbot that knows your actual academic profile.

Features

Course Plan — Semester-by-semester drag-and-drop planning board. Add and remove courses, move them between semesters, and see prerequisite conflicts highlighted in real time. Plan persists to Firebase per user account.
DegreeWorks View — Full audit dashboard: completed requirements, in-progress courses, GPA, units applied, and requirement block status — all pulled from real DegreeWorks data.
Degree Progress — Live progress bars for Core CS, Mathematics, GE, and Science requirements, powered by actual DegreeWorks data.
Explore — Browse and search 988 real UOP courses with fuzzy search, subject filtering, and credit range filtering.
AI Scheduler — Automatically generates a personalized multi-semester graduation plan based on your outstanding DegreeWorks requirements, completed courses, and target credits per semester. Uses a custom prerequisite-aware topological scheduling algorithm — no external AI call.
Tiger Advisor — Floating AI chat assistant (powered by Groq's llama-3.3-70b-versatile) that knows your full academic profile. Answers questions about prerequisites, graduation requirements, and what to take next, all grounded in your actual plan.
Export — Download your plan as a PDF or export to Google Calendar / Apple Calendar as a .ics file.

Tech Stack

Tool	Role
Next.js (App Router, TypeScript)	Framework — routing, pages, serverless API endpoints
Tailwind CSS v4	All styling and layout
Clerk	User authentication and session management
Firebase Firestore	Stores each student's course plan, keyed by user ID
Groq API (llama-3.3-70b-versatile)	Powers the Tiger Advisor AI chatbot
Custom Algorithmic Scheduler	Prerequisite-aware topological scheduling in TypeScript — no external AI
@dnd-kit	Drag-and-drop in the course plan board
Vercel	Hosting with automatic deploys on every GitHub push

How We Built It

The hackathon was 11 hours. I wrote the build guide ahead of time to map out the data strategy, tech stack, and hour-by-hour plan before we even opened our laptops. The most critical step was data collection — we used real DegreeWorks audit data and scraped course information from UOP's College Scheduler portal to populate the app with actual content.

Hours 1–2: Project setup, MyPacific-style shell layout, Clerk auth, Firebase connection
Hours 3–4: Core Plan page — SemesterCard, CourseCard, drag-and-drop with dnd-kit, Firebase persistence
Hour 5: Explore page with fuzzy search across 988 courses using Fuse.js
Hours 6–7: AI Scheduler — custom topological scheduling algorithm, Gemini → Groq switch mid-hackathon when we hit Gemini's free tier limits
Hour 8: Tiger Advisor chatbot, DegreeWorks data integration, prerequisite conflict highlighting
Hours 9–11: Export features (PDF + .ics), demo prep, Vercel deploy

The most interesting engineering challenge was the AI Scheduler. We originally planned to use Gemini to generate schedules, but the prompt → parse → validate loop was fragile. We ended up writing a deterministic topological scheduling algorithm in TypeScript instead — it reads outstanding DegreeWorks requirements, resolves prerequisite chains, and distributes courses across semesters based on the student's target credit load. More reliable, faster, and no API call needed.

The Tiger Advisor chatbot switched from Gemini to Groq mid-hackathon after hitting the free tier quota. Groq's llama-3.3-70b was faster and handled longer context better for the academic planning prompts.

What I Learned

Next.js App Router and serverless API routes in a production deployment
Clerk for auth — drop-in login flow that took under an hour to wire up
Firebase Firestore real-time database design and security rules
Groq API and prompt engineering for context-aware AI responses
Topological sorting as a practical scheduling algorithm — not just a CS concept
How to scope a hackathon project — what to cut, what to keep, and how to build for demo impact
Scraping and structuring real institutional data (DegreeWorks, College Scheduler)

Team

Alyan Anwar
Sameer Salehi
Tobias Follett

University of the Pacific — PacificHacks, April 2026