Abigail Haddad

Interactive dashboard comparing federal job listings across 2024 and 2025. Bubble charts show year-over-year changes by department, agency, and occupation — which turns out to require surprisingly fiddly math when bubbles overlap. Includes downloadable raw data for deeper analysis.

Automated pipeline collecting 3M+ federal job postings, deduplicated across two APIs that don't agree with each other. Playwright scrapes the job questionnaires that the API doesn't provide, and GitHub Actions keeps it running daily without me touching it.

Started as a one-off analysis of 35K Schedule F comments, then I realized the pattern was reusable. LLMs discover stances first, then classify at scale — which means the framework adapts to new topics without rewriting classification rules.

Cleaned up 27 years of OPM FedScope workforce data so it's actually usable — 150M+ rows that originally required joining a dozen separate lookup tables by hand. Pre-joined and standardized so you can start analyzing instead of data-wrangling.

Pick any number, count the letters in its name, repeat. You always end up at four. Interactive visualization of why this works, with flow graphs and transition tables.

A grid of hidden numbers, a secret math rule, and five lives. Figure out which numbers are safe to step on before you run out of chances. Eight levels of increasingly tricky rules.

A branching-narrative game set during Gravity Falls: Weirdmageddon. Four paths, dozens of endings (most of them bad), and Bill Cipher speaks in ALL CAPS. Styled like Journal 3.