Every candidate claim traces back to a source.
The evidence engine links claims to resume pages, snippets, parser confidence, scorecard versions, and reviewer actions.
DPA supportSSO / SCIMAudit exportsNo model training by default
ClaimLed platform team of 5 engineers94% confidence
Sourceresume.pdf · p.2led 5 engineers
DecisionShortlist reviewedScorecard v3.1
Product proof
Every score includes its evidence.
CVault’s visual grammar is built around source chips, page references, confidence markers, audit rails, and evidence export packages.
- Source-aware scoring
- Page-level provenance
- Confidence you can trust
- Audit-ready by design
Owned delivery for a real-time payments platform and led 5 engineers across backend services. Designed APIs for transaction ingestion, monitoring, and customer-facing reliability.
{
"candidate_id": "cand_789",
"role_id": "role_123",
"role_fit": 0.94,
"parser_confidence": 0.98,
"evidence_points": 28,
"review_state": "in_review",
"ats_status": "synced"
}Evidence scoring
Generic AI scoring versus CVault evidence scoring.
The difference is not prettier summarization. The difference is traceability.
Produces a confident summary, but hides the source path, confidence, reviewer state, and exportable audit evidence.
Preserves page references, document snippets, parser version, scorecard criteria, reviewer actions, and ATS handoff state.
Private pilot
Run a sample roleTalk to salesRun one sample role through CVault.
Use permitted resumes, one live role, private workspace controls, and exportable evidence before a wider rollout.