13 hours ago
London, United KingdomStaff+
Responsibilities
- Design and implement detection strategies for AI-specific attacks across distributed infrastructure.
- Build automated response playbooks, orchestration workflows, and remediation capabilities for compromised inference endpoints.
- Lead security incident response, forensic investigations, and incident communications across Cloud, AppSec, Enterprise, and AI Security teams.
- Proactively hunt for threats across GPU clusters and training infrastructure and reproduce AI-specific vulnerabilities.
- Build detection-as-code frameworks, onboard telemetry, and create dashboards for model security, GPU utilization, and sensitive research data access.
- Translate threat research into production detections and collaborate with infrastructure, software security, and AI research teams.
- Participate in a 24/7 on-call rotation for critical AI security incidents.
Requirements
- 3-5+ years of experience in security operations, detection engineering, or incident response.
- 3+ years of experience securing AI/ML infrastructure, high-performance computing environments, or distributed systems at scale.
- Strong programming skills in Python, KQL, SPL, or similar languages for detection logic, response automation, and security tooling.
- Experience with SIEM platforms, detection technologies, and forensic investigation techniques.
- Demonstrated ability to detect novel attack techniques, investigate complex distributed environments, secure high-value intellectual property, automate incident response, and conduct proactive threat hunting.
Benefits
- Generous paid time off and company holidays for UK full-time employees.
- Comprehensive medical and dental insurance.
- Paid parental leave of 16 weeks for all parents.
- Fertility and family planning support.
- Early-detection cancer testing through Galleri.
- Competitive pension scheme and company contribution.
- Wellness, learning and development, and work-life stipends.
- Company-wide and team off-sites.
- Company stock options.
