Back to work
PROJECT / 004
CybersecurityAI/ML
Advanced Multi-Heuristic Phishing Detection Platform
TLS, WHOIS, OCR, and CLIP visual similarity — fused into one trust score.
PythonFlaskTLS/WHOIS AnalysisOCRCLIP Visual SimilarityDecision TreeRandom ForestXGBoost
View repository Overview
A Flask-based multi-heuristic phishing detection platform that classifies URLs as Safe, Suspicious, or Phishing by combining certificate, domain, and visual-similarity signals.
My Role
Designed and built the detection platform; presented the accompanying research.
Status
Completed
The Problem
Single-signal phishing detectors (blacklists, or URL-string heuristics alone) miss well-crafted lookalike pages and fresh domains that haven't been blacklisted yet.
The Solution
- Combined TLS certificate validation, WHOIS domain analysis, OCR, and CLIP-based visual similarity into a single trust score.
- Presented the accompanying research, "Multi-Heuristic Phishing Detection System Using TLS, WHOIS, and OCR," at ICAISDGs 2025.
- Separately implemented and benchmarked a machine-learning classifier comparing Decision Tree, Random Forest, and XGBoost on extracted URL features.
Results & Metrics
ICAISDGs 2025
Presented at
Next project
Biometric Auth