AI Security Engineer

Key Responsibilities

Secure the AI and ML lifecycle

• Define security requirements for model development, training, evaluation, deployment, and monitoring
• Threat model AI systems and AI features in products, including abuse cases and misuse scenarios
• Establish secure-by-design patterns for model endpoints, prompts, RAG pipelines, and agent workflows
• Validate controls for model access, rate limiting, tenant isolation, and secrets management

Protect data used by AI

• Classify and control training data, fine-tuning data, prompts, and retrieved context
• Implement guardrails for sensitive data exposure, including PII and PCI data
• Define retention, deletion, and lineage requirements for AI datasets and outputs
• Partner with Privacy and Legal on data handling, regulatory expectations, and third-party data use

Secure AI infrastructure and supply chain

• Harden AI platforms, GPU and container workloads, model registries, and artifact stores
• Assess risks in third-party models, libraries, embeddings, and model hosting services
• Define integrity controls for model artifacts, evaluation sets, and pipeline automation
• Build CI and CD checks for AI assets, including scanning, signing, and policy enforcement

Detection, monitoring, and response for AI threats

• Build logging standards for model requests, responses, tool calls, and retrieval events
• Create detections for prompt injection, data exfiltration attempts, model extraction signals, and anomalous usage
• Develop incident response playbooks for AI events, including containment and rollback plans
• Run security testing for AI features, including red teaming and structured adversarial testing

Governance and program delivery

• Create practical AI security standards, patterns, and reference architectures
• Define KPIs such as reduction in sensitive output leakage, time to detect misuse, and policy coverage
• Lead risk reviews for new AI features and vendor assessments for AI services
• Train engineering and data science teams on secure AI patterns and common attack paths

Tools and Technologies You Might Use

• Cloud: AWS, Azure, GCP
• Containers: Kubernetes, Docker
• DevSecOps: GitHub Actions, GitLab CI, Azure DevOps, Terraform
• Security: SIEM, EDR, WAF, API gateways, secrets managers
• AI stack: model gateways, vector databases, model registries, ML pipelines

Examples of Work and Technical Scope

• Secure an LLM gateway with authentication, authorization, quotas, content filtering, and audit logging
• Add prompt injection defenses for an agent that uses tools like web search and internal APIs
• Implement retrieval filtering, context redaction, and output scanning for a RAG application
• Build model artifact signing and verification into the release pipeline
• Create detections in SIEM for abnormal model usage, including model scraping patterns

Required Qualifications

• 8 years of experience in Information Security, but 5+ years in security engineering, application security, cloud security, or detection engineering if there is no degree
• At least 3 years of experience in AI Security
• Experience securing LLM-based applications, RAG systems, or agentic workflows
• Familiarity with adversarial ML concepts, such as prompt injection, model inversion, and model extraction
• Experience with one or more cloud platforms, AWS, Azure, or GCP
• Hands-on experience with at least one programming language, Python preferred
• Strong understanding of AI, LLMs, API security, identity, secrets management, and cloud controls
• Experience building security controls into CI and CD pipelines
• Proven ability to lead cross-functional security work with engineering and product teams
• Effectively communicate complex technical concepts to both technical and non-technical stakeholders
• Effectively communicate to leadership and know when to escalate with proactive, clear, data-driven insight, highlighting risks, roadblocks, and solutions
• Proven leadership capabilities with the ability to influence and drive change