GPU & AI Security Researcher — Production Detections

Stealthium; United States; Remote; USD 150,000 - 190,000; Full time

Job summary

Stealthium is seeking a research-oriented security engineer to explore the evolving attack surface around GPU and AI infrastructure. The role blends vulnerability discovery with building production-grade runtime detections to guard AI workloads and accelerated compute platforms.

You will operate at the boundary between offensive security research and defensive platform engineering, producing actionable write-ups, remediation guidance, and threat intel for engineering and security teams.

Qualifications

• Proven experience in security research, vulnerability discovery, or offensive security.
• Strong knowledge of GPU compute stacks, AI runtimes, or low-level system internals.
• Familiarity with containerization, virtualization, and multi-tenant security models.
• Ability to write production-quality code for detection, telemetry, and instrumentation systems.
• Understanding of model supply chain risks and AI threat models.

Responsibilities

• Conduct original security research across GPU drivers, runtimes, and AI infrastructure.
• Investigate container and virtualization boundaries for tenant isolation weaknesses.
• Audit model supply chains for integrity, provenance, and trust gaps.
• Design and ship runtime detections that convert findings into production controls.
• Produce technical write-ups and remediation guidance for engineering and platform teams.
• Track emerging threats and disclosures in AI and accelerated-compute environments.

Skills

Security research

Vulnerability discovery

GPU compute stacks

Containerization

Production-quality code

Threat modeling

Job description

Stealthium is seeking a research-oriented security engineer to explore the evolving attack surface around GPU and AI infrastructure. The role blends vulnerability discovery with building production-grade runtime detections to guard AI workloads and accelerated compute platforms.

You will operate at the boundary between offensive security research and defensive platform engineering, producing actionable write-ups, remediation guidance, and threat intel for engineering and security teams.