Senior AI Security Engineer, Agentic Platform Guardrails

United States Digital Space LLC; Washington (District of Columbia); Hybrid; USD 180,000 - 248,000; Full time

Benefits offered by this job

Equity

Bonus

Health insurance

Dental insurance

Vision insurance

401(k)

Paid leave

Job summary

United States Digital Space LLC is seeking a Staff AI Security Engineer to define the technical vision, architecture, and controls for AI models and autonomous systems. This senior IC will mentor engineers and partner with cross-functional teams to build secure paved roads for AI innovation.

You will lead threat modeling, security reviews, and the development of enterprise security platforms, while guiding engineering velocity with developer-first tools.

Qualifications

• Deep knowledge of AI threat landscape and related frameworks (OWASP, MITRE ATLAS, NIST AI RMF).
• 8+ years of security engineering or backend software development experience with Python/Go.
• Experience securing agentic frameworks and sandboxed environments.

Responsibilities

• Design and deploy enterprise AI guardrails, gateways, and controls for agentic workflows.
• Establish security architectures for local and hosted agents to mitigate tool calling and prompt injection risks.
• Develop secure design patterns and developer tools to enable safe AI feature deployment.

Skills

AI Security

Software Engineering

Agentic Frameworks

Cloud Infrastructure

Security Architecture

Technical Leadership

Education

Bachelor's degree in Computer Science / Cybersecurity

Tools

LangChain

Strands

Claude Agent SDK

AWS

GCP

Job description

United States Digital Space LLC is seeking a Staff AI Security Engineer to define the technical vision, architecture, and controls for AI models and autonomous systems. This senior IC will mentor engineers and partner with cross-functional teams to build secure paved roads for AI innovation.

You will lead threat modeling, security reviews, and the development of enterprise security platforms, while guiding engineering velocity with developer-first tools.