How scoring works

The three scores

• Profile Score measures your public GitHub work: how much you have built, how deeply you have worked on your main projects, how consistently, the range of your stack and domains, your merged contributions to other people's projects, and the architecture of your most substantial repositories. Private repositories and work done inside companies don't appear on GitHub, so they aren't in this score. A thin GitHub history is shown as “limited evidence”, not as weak.
• Match Score measures how closely your public work lines up with a specific job's skills and domain.
• Project Score, for jobs that set a project, measures the work you submit against the project's milestones.

The three scores are kept separate. They are never blended into a single number.

What stays out of the scores

• Your name and photo. We don't ask for your gender, age, caste, religion or similar details at all.
• What you tell us yourself: your story, work history, listed skills, education and references. Companies see these separately, labelled as provided by you.
• Popularity. Stars and forks never lower a score. A widely used project can add a small, capped bonus, and merged work in an established project counts for a little more than in an unknown one.

Your stated years of experience are used in two limited ways: to pick which weighting applies to you (early career, mid-level or senior), and, for senior roles, to compare against the experience the job asks for. Where they appear they are marked as unverified.

How suggestions are ordered

Separately from the scores, Colrnx suggests roles to candidates and candidates to companies. These suggestions use the domains and skills you said you want to work in, your stated years of experience, and whether your location suits the role's work mode, alongside your public work. They only decide what is suggested first. They don't change your scores or a company's recommendation about you.

Where AI is used

• One part of the Profile Score, architecture, is assessed by an AI model reading your repositories' descriptions and READMEs. The other parts are fixed formulas.
• AI turns a job description into the skills it needs, and finds candidates whose work is relevant.
• AI writes the plain-language explanations companies read alongside the scores.

The recommendation a company sees (“strong yes”, “worth a look”, “likely no” or “not enough signal”) is decided by fixed rules over the scores, not by an AI model. The same inputs always give the same recommendation, and every recommendation lists the factors behind it.

People make the decisions

Colrnx helps companies choose whom to interview. It doesn't hire or reject anyone. Our Terms require companies to make their own decisions and to follow employment law, including rules on automated hiring tools.

If you think an assessment is wrong

You can ask for a person at Colrnx to review your assessment, and to correct any data about you that is wrong. See “Your rights” in the Privacy Policy for how to ask.