Staff Security Engineer, AI Security Labs

Anthropic; New York (NY); On-site; USD 180,000 - 260,000; Full time

Benefits offered by this job

Comprehensive health, dental, and vis

Fertility benefits via Carrot Fertilty

22 weeks of paid parental leave

Flexible paid time off

Mental health support

Relocation support

Job summary

Anthropic seeks a security-oriented engineer to lead Security Labs projects end to end, from scoping to prototype, testing against real workloads and delivering actionable writeups. You will work with researchers and infrastructure teams to validate workflows and drive secure, high-assurance systems.

You’ll combine deep cryptographic thinking with practical engineering across firmware, OS and hypervisor domains, aiming to harden AI systems against sophisticated threats.

Qualifications

• Bachelor’s degree or equivalent experience is expected.
• Experience in security research and prototyping across systems is essential.
• Strong programming ability in multiple languages and deep technical breadth required.

Responsibilities

• Own Security Labs projects end to end from scope to prototype
• Run experiments against real workloads and document results
• Create short writeups translating results into security architecture decisions
• Collaborate with research and infrastructure teams to validate workflows

Skills

Software engineering

Firmware security

Applied cryptography

OS / kernel / hypervisor internals

Formal methods

Reverse engineering

High-assurance systems

Python

Rust

Go

C/C++

Education

Bachelor’s degree

Tools

Python programming

Rust

Go

C/C++

Formal verification tools

Job description

Anthropic seeks a security-oriented engineer to lead Security Labs projects end to end, from scoping to prototype, testing against real workloads and delivering actionable writeups. You will work with researchers and infrastructure teams to validate workflows and drive secure, high-assurance systems.

You’ll combine deep cryptographic thinking with practical engineering across firmware, OS and hypervisor domains, aiming to harden AI systems against sophisticated threats.