Research Assistant - 1st qualification period (PhD candidate)

Technische Universität Berlin

Faculty IV - Electrical Engineering and Computer Science, Institute of Telecommunication Systems / Distributed and Operating Systems

Research Assistant - 1st qualification period (PhD candidate)

part-time employment may be possible

Your responsibility

Research and teaching tasks in the research group Distributed Operating Systems (DOS). Development and experimental evaluation of methods for reliability of Al infrastructures, distributed Al training and inference, energy efficiency and performance optimization for Al infrastructures. Further development of our observability tools for computing infrastructures and energy systems to analyze and evaluate the behavior and sustainability impacts of large-scale Al deployments. Publication of research results in international conferences.

Your profile

• Successfully completed university degree (Master, Diplom or equivalent) in Computer Science or a closely related field
• Solid research experience in Al infrastructure management, energy-aware computing, or the resource-efficient operation of large-scale computing infrastructures
• Strong background in machine learning and artificial intelligence, with particular emphasis on Al systems for training and inference at scale
• Experience with performance, efficiency, and resource modeling of modern Al workloads, including accelerator-based computing and large-scale model deployment
• Expertise in data-driven modeling, forecasting, and optimization methods for evaluating and controlling system behavior
• Excellent programming skills in Python and experience with scientific computing, machine learning frameworks (e.g., PyTorch), and optimization tools
• The ability to teach in German and/or in English is required; willingness to acquire the respective missing language skills.
• Experience with scalable data analysis, distributed systems, and cloud or data center infrastructures. - Familiarity with reliability and fault tolerance for Al inference and training is highly desirable