Security Engineer with Agent Authorization
First seen by Alion on Sep 30, 2026.
Technology stack
Cedar policy engine, OAuth, API keys, IAM roles, JWT, AWS Secrets Manager, AWS Bedrock AgentCore, OPA, MCP
Project overview
You will contribute to an AI platform centered on secure agent operations and controlled tool access. The work includes authorization, credential management, identity validation, audit tracing, and gateway based protection for tool interactions.
Position overview
We are looking for a Security Engineer to support the security of an AI platform with a focus on agent authorization, credential management, identity, and observability. In this role, you will work on practical security controls that help protect agent workflows and tool integrations.
Responsibilities
Implement authorization systems for API and tool access
Manage credential storage and secret rotation patterns for target systems
Validate agent identity and support cross agent authentication and service to service trust
Improve observability and tracing for agent workflows and tool invocation activity
Support audit logging for agent interactions and tool access events
Enforce gateway controls to prevent direct tool access outside approved paths
Work with AWS based secrets management for per target credential storage and rotation
Requirements
3+ years of security engineering experience in AI or ML platforms or agent platforms
Experience implementing authorization systems for API or tool access
Experience with credential management and secret rotation patterns
Experience with agent workflow observability and tracing
Knowledge of Cedar policy engine for agent to tool permissions and tool level access control
Experience with OAuth tokens, API keys, and IAM roles for secure credential handling
Experience validating JWT and supporting service to service trust models
Experience with audit logging and tracing for agent workflows
Nice to have
Experience with Cedar policy language or OPA for agent authorization
Experience with AWS Bedrock AgentCore policy configuration
Knowledge of agent to agent authentication patterns
Experience with zero trust architecture for agentic systems
Knowledge of MCP protocol security including authentication and authorization at tool invocation