Sultan Faqihi
Penetration Tester
Buad
Penetration Tester & Cybersecurity Professional (eWPT, eWPTX) with 3+ years of experience specializing in web, mobile, and API security. Reported 200+ vulnerabilities on platforms like Bugcrowd and Intigriti using deep manual testing and OWASP standards. Recognized at BlackHat MEA and trainer to 200+ students.Founder & CEO of BUAD, building an Arabic cybersecurity hub, and current trainee at Tuwaiq Academy. Actively integrates AI into workflows to automate code review and enhance security testing efficiency.
Speaker sessions
Briefings and Arsenals | Confirmed Speakers | No Session Assigned
Placeholder session for confirmed speakers without an assigned session, allowing them to be pushed to Swapcard and Visit.
- 07:30
- Tue
- 01 Dec
Stage:
Briefing Stage 2
Sessions Type:
Keynote
Breaking the 100ms Barrier: Real-Time Explainable AI for Proactive Zero-Day IoT Protection
Resource-constrained Internet of Things (IoT) edge devices face severe vulnerabilities against zero-day exploits due to their computational limitations. While Deep Learning (DL) Intrusion Detection Systems (IDS) offer high detection rates, their "black-box" nature lacks transparency. Traditional Explainable AI (XAI) frameworks (such as SHAP and LIME) incur significant response latency on edge hardware�often exceeding 170ms to 223ms�making real-time mitigation impossible and converting security responses into reactive post-incident measures.This session presents a novel proactive edge-defense framework that breaks the computational explanation barrier for resource-constrained IoT environments. We introduce a dual-quantization methodology applied simultaneously to a one-class Autoencoder-DNN model (ASAE-DNN) and the internal evaluation loop of the LIME explainability unit. By reducing computational precision from 32-bit to 8-bit, our architecture achieves an aggressive 74% reduction in model size while retaining a 97.35% detection recall against unknown zero-day attack vectors.Crucially, the dual-quantized XAI unit lowers explanation latency to 88ms, crossing the critical 100ms proactive operation threshold. This enables feature-guided, automated device isolation driven by instant, trustworthy evidence before system compromise occurs. We evaluate explanation reliability using quantitative faithfulness (92%) and robustness metrics, proving that edge-level transparency and ultra-low latency can co-exist effectively.
- 17:30
- Thu
- 03 Dec
Stage:
Briefing Stage 2
Sessions Type:
Keynote