An AI-built job search pipeline,
shipped end-to-end.
Job Engine Web collects postings from Gmail alerts and the Adzuna API, scores fit against my profile with Claude, and produces a tailored resume per role — straight into a Google Doc.
Spec-driven and pair-programmed with Claude Code. I owned the architecture, schema, and review; the implementation was authored under direction.
Five stages, fully autonomous between collection and final review.
- 01CollectGitHub Actions · 8×/dayGmail alerts + Adzuna API → Supabase
- 02RankClaude Haiku 4.5Score fit against profile skills
- 03TailorClaude Sonnet 4.6[ORIGINAL] → [REVISED] diff pairs
- 04ApplyGoogle Docs APIbatchUpdate replaceAllText
- 05TrackSupabase + SheetsStatus, notes, response cycle
- 01CollectGitHub Actions · 8×/dayGmail alerts + Adzuna API → Supabase
- 02RankClaude Haiku 4.5Score fit against profile skills
- 03TailorClaude Sonnet 4.6[ORIGINAL] → [REVISED] diff pairs
- 04ApplyGoogle Docs APIbatchUpdate replaceAllText
- 05TrackSupabase + SheetsStatus, notes, response cycle
What it actually does.
Every card below is a working surface in the app. The numbers and company names shown here are placeholders — the real pipeline stays behind the sign-in.
- Python + SQL match
- Tableau dashboard experience
- Stakeholder reporting
via Python + Airflow on AWS, surfaced
in Tableau for stakeholder review
Installable on iOS. Triage and tailor on the go — review the diff, open the Doc, mark applied.
Standard stack, spec-driven workflow.
No exotic frameworks. The interesting part is the workflow: every feature starts as a spec in PRODUCT_SPEC.md, then Claude Code authors the implementation under direction. I own the architecture, schema, and review.
Four surfaces behind the sign-in.
Rendered mockups using the live dashboard's palette — no real data behind the sign-in is shown.
Scroll →