CyberShield — AI-powered security scanning platform (graduation project)
A web platform that scans links, files, images and text for phishing and attacks, with an explainable result tailored to each type of user and an in-platform chat assistant. Built as a university graduation project in information technology.
- Client
- CyberShield
- Services
- Website interface design · Business software & POS systems
The challenge
Existing security tools are complex and fragmented, detect threats without clear explanations, and give every user the same result. The goal: one simple platform that explains the risk and recommends what to do.
What I did
• Link scanning: typosquatting, homograph attacks and suspicious domains.
• Injection detection: SQL, XSS, command injection, SSRF, path traversal, LDAP and XXE.
• Image scanning (EXIF and location data, steganography, malicious SVG), archives (zip bombs) and files.
• Text analysis for phishing and social engineering.
• A RAG engine built on OWASP rules, past cases and a knowledge base, with an explanation layer that presents results by user role.
• The in-platform CyberShield AI chat assistant.
• Admin panel for users, scans, reports, logs and settings, and a user panel with CSV and Excel export.
• Account security: hashed passwords, CSRF protection, lockout after failed attempts and session timeout.
The solution
A dark interface with neon-green accents — Arabic for users, English for the admin — built with PHP, MySQL, Bootstrap and Chart.js. Every scan returns a risk score, the main reason, an explanation and recommendations, presented differently for general, business, university and admin users. Analysis runs on rules and knowledge-base retrieval (RAG) inside the platform itself, without sending files to an outside service.
