JobShield — AI-Powered Work Opportunity Verification

Stop recruitment scams before they waste your time—AI verification in 3 seconds

JobShield analyzes 47 fraud signals across job postings, recruiter profiles, and company registration data using graph-based pattern detection and TensorFlow text classification. Our browser extension flags suspicious postings on LinkedIn and Indeed in real-time, protecting remote workers from the surge in fake recruiter scams that cost job seekers an average of $2,000 and 40 hours per incident. We've identified that 23% of remote job postings contain at least one fraud indicator—JobShield catches them before you click apply.

Key Benefits:

- Real-time fraud scoring on job postings with Neo4j graph analysis detecting recruiter network patterns and shell company clusters

- Browser extension integration with LinkedIn, Indeed, and email that flags suspicious communication metadata and posting anomalies instantly

- Company registration cross-verification against government databases, catching 89% of fake employer profiles using non-existent business entities

MVP Scope: MVP includes a browser extension that analyzes job postings for fraud signals using text analysis and company verification. Core features: real-time risk scoring (0-100), red flag identification (urgency markers, payment requests, domain reputation checks), recruiter profile analysis, and a basic dashboard showing verification results. Integration with 2-3 company registration APIs and email domain reputation services. Supports LinkedIn and Indeed job postings.

Tech Stack: Python/FastAPI for ML backend, TensorFlow for text classification, Neo4j for graph analysis, PostgreSQL for data storage, React for dashboard UI, Chrome extension API, REST APIs for third-party integrations

Components:

- Multi-signal fraud detection pipeline

- Browser extension for job posting verification

- Company registration verification engine

- Recruiter profile analysis system

- Real-time risk scoring dashboard

Quality assessment: Strong product-market fit addressing a real, quantified problem (job scam losses) with specific technical implementation (47 fraud signals, graph analysis, TensorFlow classification), but lacks originality in approach—fraud detection pipelines and browser extensions are established patterns—and the artifact is incomplete (truncated pitch and architecture sections limit depth assessment).