CheriHire

Evidence-first screening

Help & getting started

Employer workflow

1 - Create a job
Dashboard - paste the job description - Create. Then open it and click Normalize job to turn it into structured requirements.
2 - Add resumes
On the job, drag-drop PDF/DOCX/TXT, or connect a folder / email mailbox under Sources. Each resume is extracted, normalized, scored, and checked for authenticity & fit.
3 - Review the ranking
Candidates are ranked with a match band, shortlist probability, authenticity and fit. Expand a row for evidence, the prediction panel, and to send feedback.
4 - Decide
Shortlist / Hold / Reject. These decisions become training labels for the model and drive per-JD KPIs.
5 - Analytics & models
Analytics shows the funnel, prediction distributions and forecasts. Models lets you train and (after review) promote an XGBoost model from shadow to active.

Job seekers (Careers)

Go to /careers, register under an employer, upload your resume to get a health score, then apply to open roles to see your match, shortlist chance, and a path to become a strong match. Track every application and see which recruiters viewed you.

Roles & access

PO / admin platform-wide scorecard.Recruiter jobs, screening, analytics, models.Employee careers portal only.

Demo logins (change them): admin/admin123, recruiter/demo123.

FAQ

No Ollama? Everything still runs on an offline extractor; install Ollama + llama3.1 for production-quality parsing.

Why "shadow" models? A trustworthy model needs real labelled outcomes. Until then, predictions are heuristic and models train in shadow until a human promotes them.

Is bias controlled? Protected attributes are never used in scoring. Optional, consented demographics power the fairness (4/5ths) report only.

Still stuck? See Support.

Help

Step-by-step workflow and FAQ. The round ? button gives help specific to whichever page you're on.

Need a person? Visit Support or read Help.