Resume and application pipeline

Structured career data in, a tailored resume out — and a test suite that checks the numbers on it are still true.

A job ad arrives as a forwarded message.

The intake bot reads it for what it actually requires, extracts the employer's AI-usage policy, and files it against every application already being tracked.
Every must-have is matched to evidence.

Not to adjectives. Each requirement is checked against a base of evidence bullets, and the ones with nothing behind them are named as gaps rather than written around.
The numbers get re-checked by running the tests they came from.

A claim guard executes each source repo's own suite. It has caught claims drifting both ways — understated after a project grew, overstated after a refactor deleted the code.

career · board

Halcyon Systems — fit report ready 91%

Northwind Rail — pack has MISSING answers 92%

Cobalt Health — scanned 3d ago, not applied 88%

Meridian Data — draft sections unedited 74%

fit scored against 31 evidence bulletsresume claims re-checked against source-repo tests rescan 06:00

job ad · forwarded

Halcyon Systems · staff engineer

must-haves matched 7 of 8

AI-usage policy extracted

01 Built a career-data pipeline that generates an ATS-safe resume PDF and LinkedIn copy from structured YAML, with a per-application overlay engine that re-orders and prunes content against a specific job description without forking the source data
02 Built an executable claim guard, inside a 380+ test suite, that runs each source repo's own tests to check the numbers stated on the resume — catching claims that drifted in both directions, understated after a project grew and overstated after a refactor deleted code
03 Built a Telegram intake bot that turns a forwarded job ad into a graded fit report — every must-have requirement matched against an evidence base, employer AI-usage policy extracted, and a verdict — feeding a served dashboard that tracks each application from scan to outcome
04 Diagnosed and fixed a silent data-loss bug in a git-history harvester that was truncating the oldest commits from catch-up scans, then built a backfill mode that recovered history that had been dropped by up to 37%.

Interface recreated with invented data — no production screenshot is published. Private system; walkthrough and code review available on request.

A job ad arrives as a forwarded message.