Note: The job is a remote job and is open to candidates in USA. RED Global is seeking an AI Security Architecture & Administration Lead to design, implement, govern, and administer security controls across an enterprise Agentic AI ecosystem, particularly Microsoft Copilot and related platforms. The role provides security architecture guidance, manages platform controls and risks, and supports secure development and governance throughout the lifecycle of AI solutions.
Responsibilities
- Define and maintain security architecture standards, patterns, and reference designs for Microsoft Copilot, Agentic AI, and AI-powered business solutions
- Design secure architectures for AI agents, orchestration layers, integrations, APIs, plugins, connectors, and knowledge repositories
- Conduct security architecture reviews for AI initiatives, ensuring alignment with enterprise security and governance standards
- Develop security controls for Retrieval Augmented Generation (RAG), agent-to-agent interactions, and AI workflow orchestration
- Establish security design principles for identity management, data protection, model access, and agent permissions
- Collaborate with Enterprise Architecture and Platform Engineering teams to embed secure-by-design principles into AI platforms
- Administer security controls across Microsoft Copilot, Copilot Studio, Azure AI Services, Microsoft 365, Entra ID, Purview, Defender, and Power Platform environments
- Configure and maintain AI security policies, access controls, tenant configurations, and governance frameworks
- Manage role-based access controls (RBAC), Conditional Access, Privileged Identity Management (PIM), and privileged access processes
- Review and approve AI agent deployments, connectors, plugins, and external integrations
- Monitor AI platform security posture and ensure compliance with approved standards
- Support platform upgrades, security enhancements, and implementation of emerging Microsoft AI security capabilities
- Integrate and optimise SAST, DAST, SCA, Infrastructure-as-Code scanning, and API security testing within AI development pipelines
- Establish security controls across AI and agentic application development lifecycles
- Develop secure SDLC standards for Copilot extensions, AI agents, APIs, automation solutions, and custom AI applications
- Partner with engineering teams to embed application security controls into CI/CD pipelines
- Define security requirements and guardrails for AI developers, makers, and platform engineers
- Lead threat modelling exercises for AI platforms, agentic architectures, and LLM-powered solutions
- Evaluate risks using industry frameworks including:
- OWASP Top 10
- OWASP API Security Top 10
- OWASP LLM Top 10
- OWASP Agentic AI Security Top 10
- MITRE ATLAS
- Identify and mitigate AI-specific risks including:
- Prompt injection
- Agent manipulation
- Excessive agency
- Tool misuse
- Sensitive data disclosure
- Model poisoning
- Training data leakage
- Indirect prompt attacks
- Supply chain compromise
- Credential abuse
- Conduct risk assessments and provide security recommendations for new AI initiatives
- Define vulnerability management standards and remediation SLAs for AI platforms and applications
- Conduct security assessments, penetration testing coordination, and vulnerability reviews
- Analyse vulnerability findings and work with engineering teams to prioritise remediation activities
- Establish risk-based prioritisation methodologies for AI and application security findings
- Monitor security posture and recommend continuous security improvements
- Implement security controls leveraging:
- Microsoft Entra ID
- Microsoft Purview
- Microsoft Defender
- Microsoft Sentinel
- Microsoft Information Protection
- Data Loss Prevention (DLP)
- Ensure appropriate governance of enterprise data exposed to AI solutions
- Implement classification, retention, access review, and information protection controls
- Support AI governance initiatives and responsible AI frameworks
- Ensure compliance with internal policies and applicable regulatory requirements
- Establish Software Bill of Materials (SBOM) controls across AI and software assets
- Assess security risks associated with:
- Foundation models
- Open-source AI frameworks
- Plugins and connectors
- Third-party AI services
- Agent marketplaces
- Implement controls for model provenance, dependency validation, and software supply chain security
- Support vendor security assessments and third-party risk reviews
- Develop monitoring strategies for AI platforms and agent activity
- Work with SOC and Security Operations teams to identify AI-related attack patterns
- Create detection use cases for AI misuse and anomalous agent behaviour
- Investigate security incidents involving Copilot, AI agents, and AI-enabled applications
- Produce security reporting and executive-level risk metrics
- Develop security standards, policies, and operating procedures for AI adoption
- Provide technical guidance and consultation to AI programme teams
- Deliver security awareness and secure AI development training
- Participate in governance boards, design authorities, and change review processes
- Serve as the primary security advisor for enterprise AI initiatives
Skills
- 7+ years of experience in Application Security, Cyber Security Engineering, Security Architecture, DevSecOps, or Platform Security
- 3+ years of experience supporting cloud-native platforms within Microsoft Azure and Microsoft 365 environments
- Experience designing secure enterprise architectures
- Experience administering Microsoft security platforms and identity services
- Experience performing threat modelling and security architecture reviews
- Experience implementing vulnerability management programmes
- Experience securing SaaS, cloud, API, and AI-enabled solutions
- Experience collaborating with architecture, engineering, and operations teams in large enterprise environments
Company Overview