Risk-based hiring pipeline
Each applicant record is scored on network, phone and identity signals, linked records are merged into a persona, and the persona is assigned a final risk bucket.
Data collection
Gathered per applicant
- Application IP
- Video interview IP
- Phone number
- Address
- GitHub profile
- LinkedIn profile
Stage 1 · Record-level scoring
Network & IP signals
- IP matches known proxy or VPN operatorHigh risk
- IP owner / ASN matches suspicious org listHigh risk
- Same IP reused across multiple applicantsHigh risk
Silent Push enrichment
- Proxy / VPN detection: identify operator behind the IP
- ASN & org lookup: flag suspicious network owners
- Traffic origin search: given a provided address, look up IPs and infrastructure associated with that location
Phone & carrier signals
- Possible VOIP number+8
- Non-fixed VOIP / internet-based numberHigh risk
- Same VOIP number reused across applicantsHigh risk
Identity & reporting signals
- LinkedIn profile not presentVaries
- LinkedIn profile appears invalid or low qualityVaries
- LinkedIn content appears copied from another personHigh risk
- GitHub profile not present or no meaningful activityVaries
- GitHub profile appears copied or fabricatedHigh risk
- Reported risk, high confidence (e.g. direct recruiter flag)+20
- Reported risk, medium confidence (e.g. automated detection)+10
- Reported risk, low confidence (e.g. routine scan match)+3
- On a known blocklist+12
- Flagged during interview processHigh risk
- Reported by recruiterHigh risk
- Address reused across multiple submissionsVaries
Stage 2 · Persona-level rollup
Signals shared across linked personas
- Base scoreMax of all records
- Same address linked to another persona+12
- Same phone linked to another persona+10
- Same application IP linked to another persona+10
Final risk bucket
High≥ 45points
Medium15–44points
Low< 15points
