A degree planner that builds a student's whole path to graduation in seconds.
Built by Tanmai Kalisipudi.
A student enters a major, a minor and a pre-professional track. GradSteps builds a semester-by-semester plan that meets every requirement and shows when they will graduate. Advisors see every student's plan in one dashboard and can tell who is off track.
The first version targets the University of Pittsburgh. This repo holds the marketing
site at gradsteps.com and janitor, the internal tool that
builds training data for the requirement parser.
The landing page walks through the product the way a student and an advisor would use it. Every mock on the page is a live React component animated with Framer Motion, so the plan, the course cards and the advisor table move as you scroll.
|
Pick a major, a minor and a track like pre-med. The card fills in the total credits and an estimated graduation term. Course cards carry an AI summary of the class, its syllabus, a contact for the department and how hard past students found it. |
|
|
A running counter shows the tuition a student saves by graduating on time. The advisor dashboard lists every student with a status of on track, check in or needs attention. |
|
Cohort charts track 4-year graduation and first-year retention, with a projection for the current class. |
|
Course catalogs write enrollment rules as free text. janitor is a small Next.js app
that builds training data to parse them. It shows one rule at a time. You type the
structured version, or accept the output of the model trained so far. Every 50 or so
labels, you fine-tune again, so the model does more of the work each round.
npm install
npm run devOpen http://localhost:3000. The page uses a fixed desktop layout, so view it in a wide
window. The site needs no environment variables. It deploys to Vercel on every push to
main.