Weighted Talent Index & JD Fingerprinting
A data science project that analyses 106 job descriptions across 10 skill dimensions to build a unified view of how different roles compare — improving how we calibrate candidates against actual market demand.
Project Tags
106 Job Descriptions, No Common Language
Every organisation writes job descriptions differently. One emphasises alignment instincts, another foregrounds policy-aware engineering, a third looks for mechanistic interpretability depth. Comparing roles across these teams meant reading each JD manually, guessing at what skills actually mattered, and hoping your intuition held.
Without a shared framework, we couldn’t benchmark roles against each other or calibrate what ‘senior’ meant across different contexts. Skill evaluation was subjective and inconsistent.
A Unified View of Roles That Improved Calibration
Engineered a weighted Talent Index scoring framework and a JD fingerprinting tool that breaks job descriptions into 10 skill dimensions. 106 JDs across 13 organisations were processed, fingerprinted, and clustered.
This gave us a much more unified view of how roles compare across different teams and organisations, which directly improved how we calibrate any given candidate’s skills objectively for different types of positions. Instead of subjective reads on individual JDs, we had a structured baseline to evaluate against.