Working Student Cybersecurity - Focus on AI Security (m/f/d)

• Analyze security risks of modern AI systems, especially LLMs, agentic AI and AI-based development tools
• Assess AI security risks in the context of product and software development at Volkswagen Infotainment
• Investigate current attack techniques and vulnerabilities in AI systems, such as prompt injection, tool abuse, model manipulation and data leakage
• Analyze and assess the OWASP Top 10 for LLMs and the OWASP Agentic AI Top 10
• Investigate and assess security controls for AI systems, such as guardrails, sandboxing, isolation techniques, IAM concepts, audit logging and monitoring, and security mechanisms for agent frameworks
• Support the assessment of security architectures for agentic AI systems
• Analyze concepts relating to agentic identities, delegation of permissions and secure tool use
• Create technical documentation, proof-of-concepts and security assessments
• Support research into current developments in AI security and AI governance
• Develop small prototypes and automations using Python and Linux tools

Studies in IT/cybersecurity, computer science or a comparable subject; strong technical affinity and an interest in cybersecurity; solid knowledge of at least some of the following security topics: cryptography, TLS, certificates and PKI, authentication, authorization and identity and access management; initial practical experience with large language models (LLMs), machine learning or AI systems, AI agents and agentic AI, as well as coding assistants such as GitHub Copilot or Cursor; good knowledge of Python or Linux shell scripting; experience working with Linux systems; independent and analytical working style; good German or English skills. Desirable qualifications include knowledge of AI security concepts and techniques, experience with VS Code, Git and CI/CD pipelines, and an understanding of modern software development processes and DevOps/DevSecOps practices.