WillSponsor.me

The Problem

International students in the U.S. face a critical information gap when searching for jobs. The question “Does this company sponsor H-1B visas?” has no single, reliable answer. Students spend countless hours digging through forums, outdated spreadsheets, and conflicting information.

“I spent 3 weeks researching companies before even starting to apply. Most of the data I found was years old.” — Anonymous MBA student, UT Austin

The Solution

WillSponsor.me aggregates and surfaces sponsorship data from multiple official sources, making it instantly searchable. Built as a serverless application on Cloudflare Workers for global edge delivery.

The main dashboard — search across 200,000+ companies in milliseconds

Key Features

1. Real-time search across 200,000+ companies
2. Historical sponsorship data with trend analysis
3. Saved searches and email alerts
4. Community verification layer

Technical Architecture

The application is built on a fully serverless stack:

Edge-first architecture: Cloudflare Workers + D1 + KV for global sub-50ms responses

• Edge compute: Cloudflare Workers for sub-50ms global response times
• Data layer: Cloudflare D1 (SQLite at the edge) + KV for caching
• Search: Custom full-text search built on Workers
• Frontend: Static-first with progressive enhancement

Code Example

Here’s how the edge search function works:

export async function handleSearch(
  request: Request,
  env: Env
): Promise<Response> {
  const url = new URL(request.url);
  const query = url.searchParams.get('q') ?? '';

  // Check KV cache first
  const cacheKey = `search:${query.toLowerCase().trim()}`;
  const cached = await env.CACHE.get(cacheKey, 'json');

  if (cached) {
    return Response.json(cached, {
      headers: { 'X-Cache': 'HIT' },
    });
  }

  // Query D1 with full-text search
  const results = await env.DB.prepare(`
    SELECT company_name, city, state, visa_class,
           approval_count, denial_count
    FROM sponsorship_data
    WHERE company_name LIKE ?
    ORDER BY approval_count DESC
    LIMIT 50
  `).bind(`%${query}%`).all();

  // Cache for 1 hour
  await env.CACHE.put(cacheKey, JSON.stringify(results), {
    expirationTtl: 3600,
  });

  return Response.json(results);
}

Data Pipeline

The data pipeline runs on a scheduled Worker:

Stage	Tool	Frequency
Ingestion	Cloudflare Worker (cron)	Weekly
Parsing	Custom CSV/JSON parser	Per ingestion
Validation	Zod schemas	Per record
Storage	D1 batch insert	Per batch
Cache invalidation	KV purge	Post-insert

Results

After launch, the platform achieved:

• 10,000+ users in the first month
• Sub-50ms response times globally
• $0 infrastructure cost (Cloudflare free tier)
• 4.8/5 average user rating

What Users Say

WillSponsor.me saved me dozens of hours. I found three companies I didn’t know sponsored in my field within minutes.

---

Lessons Learned

Building this project taught me several things:

• Serverless-first architecture dramatically reduces operational overhead
• Edge computing makes geographic distribution trivial
• MVP scope discipline is crucial — ship fast, iterate faster
• Still exploring: ML-based company recommendation engine
• Planned: Integration with job board APIs

The Stack

• Originally considered AWS Lambda — switched to Cloudflare Workers for better cold start performance
• Frontend: Vanilla HTML/CSS/JS (no framework overhead)
• Backend: Cloudflare Workers + D1 + KV
• CI/CD: GitHub Actions + Wrangler

A Note on Ethics

All data used is from publicly available government sources. No scraping of private platforms. User privacy is paramount — no tracking, no ads, no data selling.

---

Built with care at UT Austin. View on GitHub | Live Demo