The challenge
Skills show up in many places and in many forms. One person lists “Python,” another writes “Python 3.x development,” and a third never lists it but has spent five years building data pipelines in Python. Relying on any single source would miss much of what people actually know.
Multiple signals, not a single source
We build each person’s skill profile from strong, direct evidence across several signals, including:
Stated skills: skills people list on their own resumes and professional profiles.
Experience: skills directly tied to a person’s specific work experiences, even when they never list those skills explicitly. For example, someone whose resume says they “built an automated demand forecasting tool using Pandas that reduced inventory costs by 18%” has Python skills, because Pandas is a Python library.
Credentials: skills included in specific certifications and credentials people have earned.
Project work: skills included in public projects and repositories people contribute to.
These are examples, not a complete list. Combining signals gives a fuller, more accurate picture than any one source, including skills people use every day but never think to write down.
A universal skills language
People describe the same capability in different ways, so every skill is standardized to one of more than 140,000 canonical skills, organized into three levels:
20 broad domains, like Technology, Healthcare, or Business & Management
200 skill clusters, like Machine Learning, Project Management, or Clinical Nursing
More than 140,000 specific skills
When you look at “data analysis,” you see everyone with that capability, however it was described or detected.
What this means for you
When you look at skill data for your graduates or target employers, you’re not seeing a raw dump of resume text. You’re seeing a cleaned, standardized, intelligence-enriched view of what your people actually know, in a language that’s consistent across every institution and employer on our platform.