About the job
We are seeking a highly skilled AI Security Research & Red team engineer to join our Red Team within the Security Threat Detection, Response and Emulation organization. This is a critical role that will be at the forefront of protecting our company and customers from malicious threats. You will be responsible for driving security research, exercises, and activities that emulate real world attackers and attacks to drive improvements in Cloudflare’s security posture focusing on AI, Agents, harnesses and LLM’s.
Responsibilities
Stay current with emerging threats and perform deep-dive research to identify AI-specific vulnerabilities and risks in addition to general vulnerabilities across Cloudflare’s products and services.
Identify AI-related attack surfaces through rigorous testing of agentic implementations and LLM usage to define requirements and implementation guidance.
Execute full-chain red team operations targeting Cloudflare’s global infrastructure, corporate networks, and product ecosystems.
Establish a rigorous framework for testing "Security Efficacy"—measuring exactly how well our WAF, EDR, and SIEM detections perform against known TTPs.
Foster a highly collaborative relationship with the Blue Team (Detection & Response) to ensure findings are translated into immediate defensive improvements.
Provide technical expertise while fostering a culture of curiosity, ethical hacking and partnering to improve Cloudflare’s security posture.
Translate complex technical exploits into risk-based narratives for leadership, helping prioritize engineering resources where they matter most.
Qualifications
Minimum
4+ years in offensive security, application security or other relevant field
Deep knowledge of AI, Coding agents, LLMs, prompt engineering, AI-related attack vectors (e.g., prompt injection, jailbreaking), and Agentic concepts
Strong background in manual penetration testing, exploit development, or cloud security (AWS/GCP/Bare Metal)
Experience using the MITRE ATT&CK framework to map coverage and identify "blind spots" in defensive telemetry
Ability to explain a complex "0-day" exploit to a non-technical stakeholder while maintaining the respect of a deep-dive engineering team
Knowledge of automated breach and attack simulation (BAS) tools, as well as custom-built frameworks for payload delivery and C2 infrastructure
Preferred
No preferred qualifications listed.