Arrive

The AI advisor every international student deserved on day one. Visa deadlines, CPT/OPT guidance, employer H-1B checks, and budget conversion. Built on live UMD data + Claude API.

Comment

Inspiration

Walking through UMD's campus during orientation week, we noticed something: international students were constantly lost — not physically, but systemically. They were Googling "how to open a bank account without SSN," missing CPT deadlines because the ISSO website is buried under 12 clicks, and converting tuition costs in their heads from rupees and yuan while trying to figure out if they could even afford groceries.

1 million international students enroll in US universities every year. None of them can access FAFSA. Most don't know their OPT application window until it's almost too late. And every single one of them arrived with the same feeling: I'm on my own here.

We built Arrive because no student should have to navigate an immigration system, a foreign financial system, and a new campus all at once — alone.

What it does

Arrive is an AI-powered command center for international students at UMD. It combines five core features:

  • AI Advisor — Answers visa, CPT/OPT, housing, banking, and tax questions using live-scraped official UMD ISSO pages, not generic AI knowledge. Every answer cites its source.

  • Deadline Tracker — Input your graduation date and Arrive calculates your exact OPT application window, CPT eligibility, and STEM extension deadlines with a live countdown dashboard.

  • H-1B Employer Lookup — Type any company name and see their real H-1B sponsorship history pulled from public USCIS data. Know before you apply.

  • Expense Manager — Pre-loaded with UMD-specific costs (SEVIS fee, SHIP insurance, College Park rent). Select your home currency — INR, CNY, NGN, BRL and 15 more — and see every cost converted in real time. Get AI budget analysis tailored to your situation.

  • DSO Email Generator — Drafts professionally worded emails to your international student office so nothing gets lost in translation.

How we built it

Frontend: Next.js 14 (App Router) + Tailwind CSS + shadcn/ui

AI: Anthropic Claude API — intent detection routes each question to the right UMD source, which gets scraped and injected into Claude's context window before responding

Scraping: Axios + Cheerio to extract main content from ISSO pages, with a 6-hour in-memory cache so we're not hammering UMD's servers

Database: Supabase for user profiles, saved expenses, and checklist state

Currency: ExchangeRate-API for real-time conversion across 17 currencies

The core architecture looks like this:

Built With

  • 14
  • anthropic
  • api
  • axios
  • cheerio
  • claude
  • compilation
  • css
  • exchangerate-api
  • for
  • identified
  • inventory
  • next.js
  • request
  • shadcn/ui
  • supabase
  • tailwind
  • technology
  • typescript
  • vercel
Share this project:

Updates