03 / 05 · automation · data_pipeline

A job search, run as a data pipeline.

Applying for jobs is a duplicated, fragmented, un-scored slog. You open six tabs, copy the same cover letter, paste the same resume, and have no idea whether you actually fit. So I turned my own search into a pipeline: scrape across multiple boards, dedupe, enrich, score against my resume with a built-in ATS tracker, and filter the rest down to what's actually worth applying to.

The pain point

A senior job search in a new country is mostly overhead. The roles live on six different boards — LinkedIn, company careers pages, Welcome to the Jungle, IrishJobs, Jobs.ie, referrals in Slack. Most of what you see doesn't fit. The same role shows up on three sites with slightly different titles. You have no quick way to know, before opening the posting, whether the role is even close to what your resume would score for.

The actual signal — "is this worth my 40 minutes writing a cover letter?" — gets lost in the noise. You end up applying to the wrong 40 jobs and missing the right 10.

What this does instead

The Job Tracker scrapes across the boards every day, deduplicates by company + title, and enriches each role with industry, posting date, visa-sponsorship flag, and a startup tag. It then scores every role against three distinct resume profiles using an ATS-style matcher — so each card shows up with a pre-computed fit number before I ever click through.

The dashboard gives me the view an operator actually needs: total roles, new today, high-priority (high ATS score + fresh), by-track breakdown (sales & GTM, product & strategy, ops & programs), and filters for visa sponsorship, startups, and freshness. An ATS Check button on each card runs a fuller match and tells me exactly which resume bullets are missing coverage for that role.

Why it's useful beyond my search

This is the pattern I'd build for any high-volume, low-signal intake problem — recruiting, BD prospecting, procurement scanning, RFP triage. The trick is the same in each: don't manually sift, score upfront with a clear rubric, let filters and thresholds do the triage. A job search is just an excuse to build the machinery.

Stack
Multi-source web scraping ATS-style resume scoring Deduplication + enrichment Three resume profiles · per-role match Filter DSL · visa · fresh · track · startup 490+ roles indexed Daily pipeline run Static HTML board · no backend