Seth Krch

Finds companies that are actually hiring, without a job board. It mines the Common Crawl URL index for public Greenhouse boards, then hits the Greenhouse API to pull every live opening — title, location, and apply link.

The interesting part is the evolution: the first pass parsed URLs by hand and looped the network calls. The optimized build replaces all of that with a single DuckDB query that reads the crawl's remote Parquet index over HTTPS (httpfs) — no download, no manual parsing — and streams results through a Rich terminal UI.

$ python optimized_JobFinder.py
✓ Found 248 index files
Searching URL Index for Greenhouse boards…
✓ Found 1,412 unique Greenhouse boards
✓ Senior Backend Engineer at acme

• Python
• Common Crawl
• DuckDB
• Requests
• Rich

A deliberately realistic medical-billing portal — built to be hard to scrape. It's a practice target for authenticated web automation, with all-synthetic data and zero real PHI.

It implements the safeguards real portals hide behind: password login, mock MFA, CSRF tokens, idle session timeout, role-based access, masked patient fields, JSON-loaded claims with cursor pagination, API rate limiting, audit logging, and single-use EOB links. The point: requests.get(page).text stops being enough.

• Python
• Flask
• SQLite
• Jinja
• Sessions / CSRF

A native desktop front-end built with PySide6 / Qt — a parameter sidebar and a sortable results table — putting a real interface on the job-search workflow instead of a scroll of terminal output.

Custom-styled widgets, a fixed-width control panel, and a responsive table view, all in hand-written Qt stylesheets. Because sometimes a person who isn't you needs to run it.

• Python
• PySide6
• Qt
• Desktop UI