"Security is not a product, but a process." — Bruce Schneier
I am a Cybersecurity Engineer specializing in Risk Management, GRC (Governance, Risk & Compliance), and AI-powered threat detection. My work bridges traditional security frameworks with cutting-edge machine learning approaches to tackle modern cyber threats.
- 🔐 Specialization: EBIOS Risk Manager | ISO 27001/27005 | NIS2 | DORA
- 🤖 Research: AI/ML for Cybersecurity — IoT Malware Detection, Anomaly Detection
- ☁️ Cloud Security: Azure Security Architecture, DevSecOps
- 📊 GRC: Risk Assessment, Compliance Automation, Security Governance
Applied EBIOS Risk Manager methodology across multiple critical sectors
| Case Study | Sector | Framework | Focus |
|---|---|---|---|
| AXA Cloud Azure | Insurance / Finance | EBIOS RM + CSP | Cloud Risk Assessment |
| Banque SPI DORA | Banking | EBIOS RM + DORA | Digital Operational Resilience |
| TransRail SupplyChain | Transport | EBIOS RM | Supply Chain Attack Scenarios |
Comparative study of supervised vs self-supervised approaches on CICIoT2023 dataset
# Key Results — IoT Malware Detection Performance
models = {
"Supervised CNN": {"accuracy": 0.982, "F1": 0.981},
"ViT Supervised": {"accuracy": 0.979, "F1": 0.977},
"Self-Supervised SimCLR": {"accuracy": 0.961, "F1": 0.958},
"Random Forest Baseline": {"accuracy": 0.945, "F1": 0.944},
}Cloud Governance, Risk & Compliance automation platform
╔══════════════════════════════════════════════════════════════╗
║ CYBERSECURITY EXPERTISE MAP ║
╠══════════════════════════════════════════════════════════════╣
║ GRC & Risk Management ║
║ ├── EBIOS Risk Manager (ANSSI) ████████████ 95% ║
║ ├── ISO 27001/27005 ███████████░ 90% ║
║ ├── DORA (Digital Operational Res.) █████████░░ 85% ║
║ └── NIS2 Directive ████████░░░ 80% ║
║ ║
║ AI for Cybersecurity ║
║ ├── Malware Detection (ML/DL) ███████████░ 90% ║
║ ├── Anomaly Detection █████████░░ 85% ║
║ ├── Vision Transformers (ViT) ████████░░░ 80% ║
║ └── Self-Supervised Learning ███████░░░░ 70% ║
║ ║
║ Cloud Security ║
║ ├── Azure Security Architecture █████████░░ 85% ║
║ ├── DevSecOps / CI-CD Security ████████░░░ 80% ║
║ └── Supply Chain Security ████████░░░ 80% ║
╚══════════════════════════════════════════════════════════════╝
Dataset: CICIoT2023 | Attack Types: DDoS, DoS, Recon, Mirai, Spoofing, Web Attacks + Benign
| Model | Accuracy | F1-Score | Approach |
|---|---|---|---|
| CNN | 98.2% | 98.1% | Supervised |
| ViT | 97.9% | 97.7% | Supervised |
| SimCLR | 96.1% | 95.8% | Self-Supervised |
| Random Forest | 94.5% | 94.4% | Baseline |
Key Insight: Self-supervised methods achieve competitive performance (~96%) without requiring fully labeled data — critical for real-world deployments where labeled IoT traffic is scarce.
"Securing the digital future, one risk assessment at a time."