/ Tools · 2026
Who Coffee Chats Whom
Interactive D3 network mapping 2,301 coffee chats among 166 PSE members at the University of Michigan across three pledge classes.

Overview
Turned a Gmail archive of coffee-chat thank-you emails into an interactive map of connections within Pi Sigma Epsilon at the University of Michigan. A Python pipeline deduplicates messages, matches member identities, and extracts one-to-one chats, then exports a limited public dataset for a static D3 visualization.
Key Features
- Force-directed network spanning the Chi, Psi, and Omega pledge classes
- Incremental Gmail Takeout ingestion with message deduplication and member matching
- Filters out workshops, panels, recruiter emails, and other non-chat messages
- Public export uses first names and last initials, with checks against leaking email addresses or message contents
- Static HTML, CSS, and JavaScript site with no backend or build step
Technologies
- D3.js
- JavaScript
- Python
- HTML
- CSS