Building ScamShield AI: An Evidence-Grounded, Multimodal Threat Intelligence System for Indian Mobile Scams
A deep dive into our architecture, SSRF-hardened sandboxing, multilingual code-mixed NLP, and current implementation progress.
🎯 Vulnerability: Server-Side Request Forgery (SSRF)
💡 Defense Fix: Enforce URL whitelisting, block private IP ranges (10.0.0.0/8, 169.254.169.254), and mandate AWS IMDSv2 headers.