Principal AI Security Expert

Overview

Principal AI Security Expert

AI Security Strategy & Architecture

• Design security architectures for AI systems including threat modeling, vulnerability assessment, and risk mitigation frameworks
• Lead security reviews and audits of AI models and systems throughout the development lifecycle
• Collaborate with product, engineering, and data science teams to integrate security into AI development processes
• Develop and maintain adversarial attack methodologies and tools for proactive security testing

Adversarial Research & Red Teaming

• Lead red teaming exercises to identify and exploit vulnerabilities in AI models and systems
• Conduct adversarial research to discover novel attack vectors against classical ML and Transformer models

Robust AI Development

• Research and implement techniques to enhance the robustness of AI models against adversarial attacks
• Evaluate and implement defenses against common attack types (evasion, poisoning, extraction, membership inference)
• Design and implement guardrails for AI systems to prevent harmful outputs and unsafe behaviors
• Design context injection prevention mechanisms and input validation frameworks
• Research and implement defenses against prompt injection, context confusion, and semantic attacks
• Implement monitoring systems to detect anomalous model behavior and potential security incidents

Education & Experience

• PhD in Computer Science, Cybersecurity, AI, or related field with demonstrated research in AI security, adversarial ML, or related areas (or Master's degree with 10+ years of professional experience)
• Minimum 10+ years of professional experience in cybersecurity, machine learning, or AI security
• **At least 8+ years in a senior or lead position with demonstrated technical leadership in security**
• **Proven track record of identifying and remediating critical security vulnerabilities in AI systems**
• Published research or significant contributions to AI security field

Technical Competencies

• Deep expertise in adversarial machine learning and attack methodologies against classical ML models, advanced knowledge of Transformer architecture vulnerabilities and attack vectors
• Hands-on experience with adversarial attack frameworks and tools (FGSM, PGD, C&W, etc.)
• Proficiency in Python and deep learning frameworks (PyTorch, TensorFlow), strong understanding of cryptography, secure computation, and privacy-preserving techniques, prompt injection, jailbreak, and context confusion attack vectors
• Proficiency with security testing tools and vulnerability assessment methodologies