Identification of web, network, and application vulnerabilities before they are exploited.
Evaluating the robustness of LLM models against prompt injection and jailbreak attacks.
Hardening of infrastructures and secure configurations (Docker, Kubernetes, AWS).
Integrating security by design into CI/CD pipelines and the development lifecycle.
Intelligent system for detection and neutralization of ransomwares based on kill chain analysis and malicious behavior. Static/dynamic analysis of executables (ELF, EXE/DLL, macOS), suspect signature detection, identification and neutralization of malicious processes with automatic quarantine. Integration of Hugging Face and traditional ML models to reduce false positives.
Digital community health platform: medical appointments, automatic medication reminders (notifications), secured digital medical record (allergies, blood type, history), geolocation of health centers. Contribution to functional and technical design with focus on medical data security.
Computer vision model for agricultural image classification (+50 classes). Uses ResNet-50 and MobileNetV2 with ML and Deep Learning approaches. Data preprocessing, model training, and performance analysis. Final accuracy: 97%.
Passive vulnerability assessment using Nmap, WhatWeb, Nikto, and OWASP ZAP. Business-oriented reporting with prioritized remediation actions.
Analysis of 3 realistic scenarios (credential phishing, BEC, internal spear phishing) with risk classification and awareness guidance.
Manual testing against OWASP API risks (BOLA, BFLA, excessive data exposure) with impact evidence and zero-trust hardening strategy.
Clean publishing structure for technical articles, risk analyses, and CTF writeups with better readability and navigation.