AstraWork
An evidence-based talent intelligence system that evaluates candidates through verifiable deliverables (Projects, GitHub, Demonstrations) instead of keyword matching.
THE BUSINESS PROBLEM
Traditional recruitment relies heavily on keyword matching in resumes, which rewards keyword stuffing rather than actual competence.
Good candidates are filtered out by ATS algorithms simply because their phrasing does not match a job description, while unqualified candidates game the system.
Recruiters drown in hundreds of unverified claims with no efficient way to evaluate tangible proof of ability.
THE IDEA
AstraWork replaces keyword screening with verifiable evidence.
It evaluates candidates based on working projects, repository activity, and demonstration artifacts, giving hiring managers genuine confidence in candidate capability.
THE BUSINESS & PROCESS ANALYSIS
I mapped the hiring funnel: Requirement Definition → Candidate Sourcing → Evidence Verification → Shortlisting.
The fundamental breakdown was at the screening stage, where resume text acts as an imperfect, easily falsified proxy for skill.
THE REQUIREMENTS & IMPLEMENTATION
Defined product requirements for candidate evidence extraction, skill verification workflows, and an employer-side configurable ranking model.
Designed configurable ranking rules (e.g., top-50 or top-200 candidate tiers) allowing employers to tune visibility and review depth according to hiring bandwidth.
THE RESULT / CURRENT STATUS
Delivered a working prototype and design system validating evidence-based hiring workflows.
Proved that structuring candidate evaluation around tangible proof cuts recruiter review fatigue and surfaces genuinely capable builders.
WHAT I LEARNED
The hard part was defining what constitutes valid skill evidence and structuring ranking algorithms that reward genuine execution over resume buzzwords.