Patent Pending
One fact. Multiple sources.
Compounding proof.
The Signal Mesh cross-references every candidate claim against 13+ independent sources. When 3 or more agree, the evidence compounds. No opinions. No keyword matching. Just deterministic corroboration.
◆ SIGNAL MESH — CORROBORATION MATRIX
● LIVE
The Problem
Resumes lie. Screening can't keep up.
AI-generated resumes are indistinguishable from real ones. Keyword matchers can't tell the difference. The only thing left is evidence — and evidence needs to be cross-referenced.
68%
of candidates admit to exaggerating their resume
400%
increase in AI-generated applications since 2024
6s
average time a recruiter spends reviewing a resume
$2K+
cost of every bad interview that shouldn't have happened
How It Works
6 fact categories. 13+ sources. One matrix.
Every claim on a resume is classified into a fact category. Each category is independently verified against multiple sources. The mesh shows exactly which sources confirmed which facts.
Fact Categories
Identity
Employment
Skills
Education
Clearance
Contributions
Independent Sources
GitHub API
LinkedIn X-ray
Email Verification
SEC EDGAR
SAM.gov
Credly + 8 more
Corroboration Tiers
The more sources agree, the higher the confidence.
Each fact category earns a tier based on how many independent sources confirm it. Three or more sources trigger the corroboration multiplier.
Unverified
0 sources. The claim exists only on the resume.
Claimed
1 source found. Present but not yet corroborated.
Confirmed
2 independent sources agree. Confidence building.
Corroborated
3+ independent sources confirm the same fact. Evidence compounds. Multiplier activates.
The Multiplier
When evidence compounds, the score rises.
Each corroborated fact adds +0.04 to the multiplier. A candidate with strong cross-referenced evidence across multiple categories can earn up to a ×1.25 boost.
×1.00
Corroboration Multiplier
finalScore = evidence × penalty × corroboration
Sources
13+ independent verification sources.
Every source is independently queried. No source trusts another. The mesh only lights up when the data matches.
👤
LinkedIn
Identity, Employment
✉
Email Verify
Deliverability Check
🏛
SAM.gov
Employment, Clearance
📚
Professional Profile
Identity, Employment, Education
🎓
Academic
Education, Publications
📰
Web Presence
Community, Contributions
🔎
Web Identity
Identity, Employment
📦
npm / PyPI
Skills, Contributions
🔐
Prof. Bodies
Certifications
Anti-Gaming
Built to resist fabrication.
Every scoring decision is deterministic. Every point traces to a verified fact. The system is designed so that faking evidence is harder than earning it.
🔒
Binary scoring
Confirmed = full points. Not confirmed = zero. No partial credit for naming real companies you didn't work at.
⚠
Fabrication detection
Dead GitHub links, fake LinkedIn profiles, and fabricated employers trigger automatic penalties and verdict overrides.
🕑
Velocity thresholds
3+ new signals in under 60 days trigger suspicion. Depth built over time scores higher than breadth created overnight.
🔗
Derived source tracking
LinkedIn found via GitHub doesn't count as independent unless the GitHub account is 12+ months old with 5+ repos. You can't bootstrap trust from a single source.
📈
Archetype-aware weights
A military officer isn't scored the same as a software engineer. Evidence categories are weighted by role type so the score reflects what matters.
◆
Patent Pending
The Signal Mesh corroboration model — including the evidence graph, corroboration multiplier, anti-gaming thresholds, and derived source tracking — is covered by provisional patent application filed by Beskar, LLC.
See the mesh in action.
Drop a resume. Watch 13+ sources light up in real time.
Book a Demo →