JobRun

Job hunting is repetitive glue work: find roles, judge fit, rewrite your CV for each one, draft a cover letter, prep for the interview, then re-type the same details into every application form. JobRun automates the glue so the applicant spends time on decisions, not data entry.

JobRun sources jobs from multiple providers and scores each against the user's profile. It builds a structured Master Profile from the user's CV, then generates tailored CVs, cover letters, and interview prep per role.

A companion Chrome extension (Manifest V3) reads the generated profile and autofills job-application forms across sites, closing the last, most tedious mile of the process.

It's a TypeScript monorepo (web app + extension + shared packages) on Next.js 16 / React 19, with Drizzle over Neon Postgres. AI generation runs through the Vercel AI SDK v6 with OpenAI. Long-running work — sourcing, scoring, document generation — is offloaded to 14 Trigger.dev background tasks so the UI stays responsive, with generated documents stored on Cloudflare R2.

Pushing generation into background jobs (rather than doing it inline) adds orchestration complexity, but it's what lets the app source and score many jobs and produce documents without blocking the user or timing out.

A monorepo spanning a web app and a browser extension means shared types and one source of truth for the profile model — at the cost of a more involved build setup.

JobRun is in active development (~241 commits) with a working live demo.