Student*in für Masterarbeit - Agentic AI in der CarIT Security

Tasks

Mercedes-Benz AG's Car IT Security department is responsible for vehicle cybersecurity throughout the product lifecycle, from threat analysis and security requirements to securing ECU software, key material, update processes, and regulatory and homologation evidence. This master's thesis addresses the use of agentic AI for knowledge-intensive engineering tasks across requirements, ticket, and documentation systems.

• Analyze the state of the art in agentic AI, including LLM-based agents, tool use, multi-agent orchestration, and retrieval of specialist knowledge, and assess applicability to safety-critical engineering processes.
• Develop an assessment framework for classifying tasks according to their suitability for agentic support, considering criticality, traceability, and required human approval points.
• Design a reference architecture for an agentic assistant connected to requirements and ticket systems, code repositories, and knowledge databases, including permissions, data concepts, guardrails, and human-in-the-loop gates.
• Implement at least one prototype use case in the existing tool environment and test it using approved specialist content.
• Evaluate the prototype quantitatively and qualitatively, including output quality, reproducibility, processing time, errors, and hallucination behavior, compared with the current manual process.
• Assess the regulatory and normative framework, including ISO/SAE 21434, UN ECE R155, the EU AI Act, and internal AI and information-security requirements.
• Develop recommendations covering a scaling path, operating model, and open risks, and present the results to the department.

The final topic will be determined in consultation with the university, the candidate, and Mercedes-Benz. The position can start in September 2026.

Qualifications

• Degree program in computer science, IT security, electrical engineering, automotive engineering, mechatronics, business informatics, data science, or a comparable field.
• Basic knowledge of IT or cybersecurity, ideally with initial exposure to automotive security or embedded systems.
• Interest in or initial practical experience with generative AI, large language models, prompting, and agent or automation frameworks.
• Programming skills, preferably Python, and confident use of common development and collaboration tools.
• Fluent German and English, written and spoken.
• Confident use of MS Office.
• Commitment, teamwork, analytical thinking, structured and independent working style, and willingness to work in a regulatory environment.

Additional information

Please submit an online application with a CV, cover letter, certificates, a current enrollment certificate stating the semester, and proof of the standard period of study. Documents must be marked as relevant to the application in the online form and must not exceed 5 MB.

Applicants with severe disabilities or equivalent status are welcome. The representative office can be contacted at sbv-sindelfingen@mercedes-benz.com. People Solutions can be reached at myhrservice@mercedes-benz.com or 0711/17-99000 for questions about the application process.