Staff Engineer, AI Security

This is a fully remote position, open to applicants in Ireland.

📋 Description

• Act as the main subject matter expert for all initiatives related to AI and machine learning security within security and R&D.

• Design and oversee AI gateways to establish a centralized control plane for authentication, authorization, and rate limiting across all interactions with models and tools.

• Create and sustain an autonomous security agentic framework that employs multi-agent orchestration for comprehensive investigation, alert triage, and remediation.

• Develop agentic identity models utilizing OAuth 2.1 to ensure identity propagation across trust boundaries and mitigate the confused deputy problem.

• Assist in governing the AI-augmented software development lifecycle by incorporating real-time security gates into the developer environment and CI/CD pipeline.

• Manage Agentic Security Solutions that safeguard the AI lifecycle and oversee AI workloads during runtime.

• Write company-wide AI security standards and apply these security measures throughout Twilio’s infrastructure.

• Implement human-in-the-loop checkpoints and transactional safety protocols for high-impact or potentially destructive agentic actions.

• Collaborate with engineering leadership to establish the long-term vision for identity-centric security and automated posture management.

• Serve as a knowledge multiplier by mentoring security engineers and creating secure-by-default paved road templates for R&D teams.

⛳️ Requirements

• Over 8 years of experience in security engineering, with a minimum of 3 years dedicated to AI or machine learning security operations (MLSecOps).

• Proficient in orchestrating multi-agent systems using AWS Strands, LangGraph, and CrewAI, with a focus on runtime isolation, PII redaction, and protection against indirect prompt injection in agentic settings.

• Practical experience with AI-specific frameworks (e.g., MITRE ATLAS, MAESTRO, OWASP Top 10 for LLMs/Agents/MCP) to threat model and defend against a wide array of risks, such as direct/indirect prompt injection, training data poisoning, tool poisoning, and data exfiltration within agentic workflows.

• Expertise in securing end-to-end AI pipelines, from data ingestion and training to model deployment and monitoring.

• Excellent communication skills to translate intricate AI risks into actionable business logic for stakeholders.

🏝️ Benefits

• Competitive pay

• Generous time off

• Ample parental and wellness leave

• Healthcare

• Retirement savings program

• And much more