AI Security Engineer

ISTARI is focused on enhancing security for AI and Machine Learning capabilities across enterprises. The AI Security Engineer will establish security controls, embed security into AI solution design, and govern the AI landscape while collaborating with various teams to ensure secure AI adoption.

Responsibilities:

• Define secure architecture patterns for AI/ML solutions, covering models, training pipelines, inference environments, and data flows
• Establish and review secure integration patterns for AI services across enterprise applications, APIs, cloud platforms, and data environments, ensuring alignment with secure-by-design principles
• Maintain hands-on familiarity with the Microsoft AI security ecosystem — Defender for AI, Microsoft Purview, and related tooling — and apply this to secure both Microsoft-native and third-party AI environments (e.g., OpenAI/ChatGPT, Anthropic/Claude)
• Develop and shape AI protection architecture end-to-end: model access controls, data governance, inference controls, and AI-specific threat mitigations
• Identify, assess, and mitigate AI-specific threats including model poisoning, prompt injection, adversarial attacks, unauthorized access, data leakage, and misuse of AI outputs
• Define and implement security guardrails for AI model access, API usage, prompt controls, and secure interaction with enterprise data sources
• Establish controls to protect sensitive training data, embeddings, prompts, and inference outputs; validate third-party AI services and external model integrations from a cybersecurity risk perspective
• Develop and own AI governance standards defining how AI systems are assessed, approved, monitored, and retired — spanning data handling, model access, acceptable use, and regulatory alignment
• Establish AI security engineering guardrails and review checkpoints for new AI initiatives, pilots, and production deployments
• Contribute to enterprise AI security policies, reference architectures, and operational standards; partner with Digital and AI teams to enable secure adoption without acting as a blocker
• Collaborate with Cyber Defense Operations to operationalize AI-related detection, monitoring, and response capabilities; define logging and telemetry requirements for AI platforms to improve visibility and incident readiness
• Partner with the DLP architect to implement AI-focused Data Loss Prevention capabilities leveraging Microsoft Purview and AI agents, preventing sensitive data exposure across AI interactions and pipelines
• Drive development of AI-based security use cases for security operations — conceiving, building, and operationalising AI-driven capabilities that deliver measurable detection, investigation, and response improvements
• Work closely with Security Architecture, Cloud Engineering, Data, Application teams, and AI program owners to ensure consistent security adoption; support security reviews for AI vendors and AI-enabled SaaS platforms
• Represent AI security in architectural review meetings and Microsoft engagements — contributing substantively to design decisions and tooling selections, challenging and defending positions with credibility

Requirements:

• 5–8 years of cybersecurity engineering or security architecture experience, with exposure to cloud security, data protection, or application security
• Hands-on experience with the Microsoft AI security ecosystem: Defender for AI, Microsoft Purview (information protection, DLP, compliance), and Microsoft Sentinel; able to configure and integrate these tools to secure third-party AI environments
• Experience working with enterprise AI/ML platforms, analytics environments, and AI/ML deployment patterns, APIs, and model lifecycle
• Demonstrated experience developing AI-based use cases for security operations — conceiving, building, and operationalising AI-driven capabilities, not just evaluating tools
• Experience developing AI governance standards and shaping AI protection architecture, including acceptable use policies, model risk assessment, and enterprise-wide AI security oversight
• Proven ability to embed secure AI controls into enterprise initiatives without impeding adoption, and to translate emerging AI risks into practical engineering controls
• Security certifications such as CISSP, CCSP, or equivalent cloud security certifications preferred
• Practical knowledge of Microsoft security tooling for AI environments: Defender for AI, Microsoft Purview (DLP, information protection, compliance), Microsoft Sentinel, and Azure AI services; able to configure, integrate, and leverage these to protect both Microsoft and third-party AI environments
• Familiarity with AI platforms, cloud-native services, secure API design, access control, encryption, and monitoring patterns
• Ability to assess and secure AI integration points across enterprise systems
• Strong understanding of cybersecurity controls across cloud, applications, APIs, identity, and data protection, including AI/ML-specific risks such as prompt injection, model abuse, and adversarial techniques
• Knowledge of secure architecture principles for modern digital and AI platforms