security engineer for AI agent platforms

EPAM develops enterprise software products, open source solutions, and accelerators. Its Enterprise Agent Development Platform enables engineering teams to define, orchestrate, deploy, and observe AI agents at scale using LangGraph and Strands Agents on AWS AgentCore Runtime, reducing agent development from months to days while enforcing security, quality, and governance standards.

• Implement the baseline Cedar policy library and policy authoring standard for the platform;
• Author and test Cedar policies governing agent access control;
• Configure AWS AgentCore Policy LOG_ONLY and ENFORCE modes for staged policy rollout;
• Build Python-based tooling to validate policy definitions;
• Map MS Entra claims to platform policy attributes;
• Design parameter-level access control patterns for agent tooling;
• Establish governance standards for policy authoring across engineering teams;
• Collaborate with platform and security teams to enforce consistent access control practices.

• 3+ Years of experience in security engineering or backend engineering;
• Expertise in Cedar policy language, including authoring and testing;
• Hands-on experience with Open Policy Agent (OPA);
• Knowledge of AWS AgentCore Policy LOG_ONLY and ENFORCE modes;
• Proficiency in Python for policy validation tooling;
• Familiarity with MS Entra claims mapping and parameter-level access control patterns;
• Skills in identity provider integration with Entra, Okta, and Cognito;
• Understanding of policy definition in an ABAC or RBAC system;
• Nice to have: familiarity with LangGraph tool invocation patterns, knowledge of AWS AgentCore Gateway integration.

• Remote work is available in Poland.