SwipeScholar

Inspiration

Most students pick PhD advisors based on research fit alone — completely ignoring that professors run labs funded by Google, NVIDIA, and AWS, and those relationships quietly determine where their students end up. We wanted to make that invisible pipeline visible, and build the tool that connects all of it.

What it does

Upload your resume, pick your target companies, swipe Tinder-style through professor cards. Each card shows research tags, match score, and which of your target companies fund that lab. Swipe right, Claude drafts a personalized cold email citing a real line from their latest paper. Resume to inbox-ready emails in under two minutes.

How we built it

Next.js 15, Tailwind, and the Anthropic SDK. Three Claude agents: a profile parser, a match scorer weighting research fit, industry funder overlap, and background fit, and an email drafter grounded in real paper content. Company preferences flow through the entire pipeline from upload to email.

Challenges we ran into

Getting Claude to cite real paper content without hallucinating — solved by passing the abstract directly into context and restricting the model to only reference what was provided. Balancing three match signals into one coherent score took most of our iteration time.

Accomplishments that we're proud of

The industry funding angle is genuinely novel. No existing tool connects lab funders to student career goals as a matching input. The full pipeline is live — nothing is mocked.

What we learned

PhD advising and career strategy are the same decision, and nobody is treating them that way. Also: multi-agent decomposition beats one giant prompt every time.

What's next for SwipeScholar

Live funding data from NSF and DARPA grant databases, Semantic Scholar for real paper abstracts, Gmail integration to send directly from the app, and alumni tracking to show where each lab's students actually end up.

Built With

• english
• next.js
• tailwindcss
• vercel