Responsibilities
- Develop and refine the enterprise AI security framework, including architectural standards, safeguards, and security criteria that support business goals.
- Integrate security-by-design principles into AI development, deployment, and operational processes, covering hardening, access management, monitoring, and testing.
- Create and implement security measures for SaaS applications enhanced with AI capabilities.
- Build security controls for internal AI agents and automated workflows.
- Secure model hosting environments, inference services, APIs, and orchestration layers through tailored security engineering.
- Design protections for retrieval-augmented generation (RAG) systems, vector databases, and embedding processes.
- Engineer safeguards for AI model training and fine-tuning pipelines.
- Develop security controls for multi-agent communication protocols and agent-to-agent interaction models.
- Extend identity and access management to cover non-human entities and autonomous AI agents.
- Treat AI agents as primary identities with defined authentication, authorization, lifecycle handling, and deactivation procedures.
- Implement delegated authorization models to differentiate actions initiated by humans versus AI agents.
- Enforce least-privilege access and scoped permissions to prevent privilege escalation in automated and multi-agent environments.
- Identify and address AI-specific threats such as data leakage, prompt injection, jailbreaking, model misuse, data poisoning, model theft, and supply chain vulnerabilities.
- Ensure security testing and validation are embedded throughout AI development and deployment lifecycles.
- Define logging, monitoring, and threat detection requirements for AI systems, models, and agent behaviors.
- Collaborate with SecOps to ensure AI-related events are visible, auditable, and can trigger responsive actions.
- Assist in responding to and analyzing security incidents involving AI systems.
- Collaborate with IAM, SecOps, AppSec, GRC, IT engineering, AI platform teams, and business units to integrate security at appropriate levels.
- Enhance enterprise data security strategies with a focus on AI-driven data access patterns.
- Deploy and optimize Microsoft Purview for data classification, sensitivity labeling, data loss prevention, information protection, and monitoring.
- Align data security controls with AI system designs to minimize exposure of sensitive data through prompts, agents, outputs, or downstream sharing.
- Support secure data usage in RAG pipelines, AI workflows, and model training environments.
- Contribute to data protection strategies across collaboration tools, cloud platforms, and endpoint devices, promoting consistent policy enforcement.
