Built for engineers who hate scrolling LinkedIn
RoleMiner is an AI-driven job search tool with a clear scope: crawl the boards engineers actually post to, score the matches against your real resume, and stop wasting your evenings.
Who's behind it
Ten years building production backends — video streaming infrastructure at FOX/X-Team, distributed exchanges on Solana and Ethereum, Rust + Go + CUDA at Kake-Cosm. The problem RoleMiner solves is the one I was solving by hand on weekends: scanning the same handful of ATS boards, re-tailoring the same CV for every application, and losing track of what I'd applied to two weeks ago.
Where the listings come from
We index 1,200+ sources: ATS systems (Greenhouse, Lever, Ashby, Workday, etc.) where companies post directly, plus the major job boards (LinkedIn, We Work Remotely, RemoteOK, and similar). All ingestion goes through Fantastic.jobs's aggregator API, which is the same data pipeline several well-known job sites use.
- The job-listing pages on this site (
/jobs/<slug>) are rebuilt every 6 hours. - The in-app live search hits the source aggregator at request time, so it's never older than the upstream cache (~10 minutes).
- Listings link out to the original posting on the company's ATS — we don't gate or paywall the application.
What we do with your data
- Resume. Stored in your Supabase row (Postgres + RLS). Used to score job fit and tailor CVs. Never shared with third parties.
- Application pipeline + memories. Same story — your row, your data, no external mirroring.
- Tailoring + scoring. Calls Anthropic's Claude API with your resume + the job description; Anthropic's commercial terms govern that path.
- No data sale, no broker resale, no ad targeting. See the privacy page for the long form.
Talk to a person
Bug? Feature you wish existed? Aggregation source you think we should add? Email me directly.
hello@roleminer.app