Boring Job
A local-first career-operations assistant that verifies jobs, matches trusted resumes and prepares review packets without submitting or sending anything.
Problem
Automated job discovery becomes unsafe when scraped pages are accepted as real vacancies or an assistant can apply and contact people without review.
How I solved it
- 01
Checks supported job-board listings deterministically, with provider fallbacks and circuit breakers, before a role reaches the review workflow.
- 02
Uses AI for resume matching and drafting inside review packets and founder-research flows, but keeps applying, messaging and any external action behind explicit human approval.
- 03
Stores deduplication, usage visibility and operator state locally so the tool remains a supervised single-operator workflow rather than an autonomous application service.
My role / what I owned
Solo builder and operator
- Built provider fallbacks, circuit breakers and deterministic checks for supported job-board listings.
- Built resume matching, review packets and founder-research flows around explicit human approval.
- Added durable deduplication, usage visibility and a local review dashboard while keeping external actions manual.
How it was built
- Team
- A self-directed project in Nivish's private repository.
- Tools
- Node.js
- Docker
- Firecrawl
- Apify
- OpenAI API
- Use of AI
- AI assists resume matching and drafting inside deterministic verification, fact and manual-review boundaries.
Source and current state
- The private history spans 42 commits from 24 July to 9 August 2026.
- Nineteen test files cover providers, discovery, verification, pipelines, stores, scheduling and the server.
- The audited README records 174 passing tests and documents every supervised external-action boundary.
- Status
- Implemented as a local, single-operator system.
- What it shows
- The audited repository demonstrates the verification, review and fallback paths without claiming autonomous application submission.