Projects

WhoDoYouKnow

Discover your real network from your Gmail history.

View on GitHub

Extracting real contacts from an email inbox and enriching them with AI.

The Story

As graduation approaches, I started thinking about all the people I've met over the last few years. Classmates, professors, hackathon teammates, random introductions—so many connections that matter to me.

I realized that if I didn't make a conscious effort to catalog these relationships, I might lose touch with many of them. I wanted a repository of the people I interacted with, somewhere I could look back and say, "Oh right, I should reach out to them."

The most honest record of my interactions wasn't a social media platform; it was my email inbox. But going through thousands of emails manually to find actual humans (and not newsletters or automated receipts) sounded awful. So I decided to build a tool to do it for me.

The Build

I hacked this together to solve my specific problem as quickly as possible. The core of WhoDoYouKnow is a Gmail scanner that reads through email threads to identify real contacts.

It filters out the noise—like one-way promotional emails—by looking for mutual communication or specific patterns. Then, it uses AI (via OpenRouter) to enrich those contacts, categorizing and summarizing the context of my relationship with each person based on our exchanges.

Privacy was a big concern for me, so I designed it to process everything in-memory. Emails are never stored permanently, and the extracted data is deleted shortly after you download your clean CSV export. It's built with Next.js, Drizzle, and Google OAuth.

Open Source

Because it relies on restricted Gmail API scopes, running a public version for everyone requires an expensive and rigorous Google security assessment. Instead, I decided to open-source the entire project.

If you want to map out your own network before graduating, changing jobs, or just for your own records, you can easily self-host it. You just need to set up a Google Cloud project in "testing" mode, which bypasses the verification requirements for your own account.

You can find the code, setup instructions, and the full self-hosting guide on GitHub.