AI Security Engineer

Crowe is a leading public accounting, consulting, and technology firm that focuses on enhancing services through innovative AI solutions. The AI Security Engineer I (Senior Staff) role involves securing enterprise AI and machine learning systems, conducting advanced security assessments, and collaborating with various teams to protect sensitive data from emerging threats.

Responsibilities:

• Architecting secure deployment and operating models for AI, ML, and generative AI systems across cloud and hybrid environments
• Conducting advanced AI security testing, including adversarial ML attacks, prompt injection simulations, and RAG manipulation assessments
• Identifying and mitigating vulnerabilities in model-serving infrastructure, feature stores, embedding pipelines, and vector databases
• Designing guardrails, safety filters, access controls, and secure interaction patterns for LLM- and RAG-based applications
• Developing automated tooling to detect misconfigurations, insecure endpoints, and data exposure risks within AI pipelines
• Collaborating with cloud and DevOps teams to secure Kubernetes clusters, GPU workloads, and infrastructure-as-code deployments
• Analyzing logs, telemetry, and model outputs to detect anomalies, abuse patterns, model degradation, or malicious activity
• Implementing encryption, secrets management, IAM policies, and network segmentation for AI workloads
• Leading secure design and architecture reviews for AI features, APIs, and platform components
• Documenting threat models, attack surfaces, risk assessments, mitigations, and compliance artifacts
• Participating in AI-specific incident response, investigation, and post-incident analysis
• Evaluating emerging AI security technologies, including model fingerprinting, inference protection, and secure execution environments
• Supporting enterprise adoption of responsible AI, data protection, and regulatory compliance standards
• Mentoring junior engineers, ML engineers, and security practitioners on AI security best practices
• Contributing to cloud security posture management capabilities for AI-enabled platforms

Requirements:

• 4+ years of experience in cybersecurity, cloud security, ML engineering, or DevSecOps roles
• Demonstrated experience securing AI/ML or generative AI systems in production environments
• Strong understanding of ML pipelines, model architectures, and AI system components
• Deep knowledge of adversarial ML attack vectors and mitigation techniques
• Proficiency in Python, security testing tools, and cloud security frameworks
• Ability to assess risk across distributed services, storage systems, inference APIs, and data pipelines
• Strong communication skills and sound technical judgment in security decision-making
• Hands-on experience with Microsoft Azure and M365 security environments
• Willingness to travel occasionally for cross-functional planning and collaboration
• Bachelor's degree in Cybersecurity, Computer Science, Engineering, or a related technical field, or equivalent experience
• Master's degree or advanced training in cybersecurity, AI, or related discipline
• Security and cloud certifications such as SC-100, SC-900, SC-200, SC-300, AZ-500, AI-102, or equivalent AWS certifications
• CISSP, CKS, or CompTIA Cloud certifications
• Advanced experience securing AI platforms on Azure, including Kubernetes security (RBAC, network policies) and multi-tenant GPU workloads
• Experience securing container pipelines using image scanning, signing, and policy enforcement
• Expertise with secrets management solutions (e.g., Azure Key Vault, HashiCorp Vault)
• Experience implementing zero-trust architecture and securing CI/CD pipelines for AI systems
• Deep knowledge of generative AI and RAG security, including prevention of prompt injections, jailbreaks, context poisoning, and embedding leakage
• Experience designing safe-output rendering patterns, guardrails, and red-teaming processes for generative systems
• Familiarity with emerging generative AI defense techniques such as model watermarking, inference integrity checks, and output validation frameworks