security engineer

EPAM provides digital platform engineering and software development services.

• Execute Direct and Indirect Prompt Injection attacks and jailbreaking techniques including multi-turn and roleplay scenarios to test LLM guardrails;
• Identify vulnerabilities related to model behavior manipulation and potential data leakage;
• Apply MITRE ATLAS and OWASP Top 10 for LLMs frameworks to structure adversarial testing methodologies;
• Test Retrieval-Augmented Generation pipelines for data poisoning vulnerabilities;
• Validate Agentic workflows against tool-use manipulation, goal hijacking and unauthorized permission escalation;
• Develop and script custom boundary-testing payloads using Python;
• Leverage AI red teaming and evaluation frameworks to assess model safety and resilience;
• Document findings and provide remediation guidance to engineering teams;
• Collaborate with cross-functional teams to strengthen defenses across AI systems.

• 5+ Years of experience in adversarial AI testing including Direct and Indirect Prompt Injection, Jailbreaking and model behavior manipulation;
• Familiarity with MITRE ATLAS, MITRE ATT&CK and OWASP Top 10 for LLMs frameworks;
• Experience testing RAG pipelines for data poisoning and validating Agentic workflows against tool-use manipulation and goal hijacking;
• Proficiency in AI red teaming tools such as PyRIT, GARAK and Giskard or enterprise suites like Confident AI or Repello;
• Understanding of LLM fundamentals including tokens, embeddings and context windows;
• Skills in Python programming for scripting custom boundary-testing payloads;
• Knowledge of LLM security, Prompt Engineering and Security practices;
• Background in Multi-Agent Systems Evaluation and Testing;
• English proficiency at B2 level or higher;
• Nice to have: Expertise in AI Solution Engineering, Knowledge of Machine Learning.

• No conditions specified