Product
JobScout AI: job discovery, CV matching, and evidence-based research
Inside an account-based job-search application with scheduled discovery, CV-based evaluation, and research that preserves sources and uncertainty.
Searching for a job involves more than collecting links. JobScout AI connects discovery, personal preferences, CV-based evaluation, and follow-up research while keeping each account’s work in one place.
Turn a CV and preferences into a search
Users import or paste their experience, review the extracted text, and work through a conversational setup. Confirmed roles, locations, and work preferences become the active search profile. Draft conversations remain separate until the settings are saved.
The worker searches supported sources such as LinkedIn, Y Combinator, Ashby, and Greenhouse. Results can be evaluated against the profile and researched further. Email notifications are optional and require a verified address.
Separate the dashboard from scheduled work
Next.js serves the account dashboard, while a TypeScript worker performs bounded search and evaluation runs. PostgreSQL stores profiles, jobs, evidence, and delivery history. The documented deployment splits the web application and worker between Vercel and DigitalOcean.
Account-scoped queries, duplicate handling, worker leases, and delivery records matter as much as the model call. Research saves checked excerpts and source dates, and checkpoints allow interrupted investigations to retain partial evidence.
A match score is a decision aid
Missing evidence remains visibly unknown. A high numeric score cannot override configured eligibility gates. Model extraction can still be wrong, and an employer-name match does not verify company identity or establish a job’s terms.
Source availability, run cadence, and AI budgets limit each search. The app does not establish hiring outcomes, and several portals and broader company-first research workflows remain outside the documented release.
Read the workflow behind the interface
The repository explains onboarding, deployment, scoring, research receipts, retries, and operational limits. It is a useful example of putting an AI feature inside an accountable product workflow rather than stopping at generated recommendations.
Sources and further reading
Based on the linked project documentation, reviewed on 2 November 2026. Capabilities and limitations describe that documented version.
View repository