PhD student in Embedded AI Systems: Runtime optimization for transformer models

You develop the state of science for scheduling AI operations, online optimization, learning-based resource allocation and real-time systems in the context of heterogeneous computing systems. From this you identify open research questions and position your dissertation. You formalize the problem of runtime orchestration (addition of limited computing/memory resources to competing tasks) as an optimization problem. To do this, analyse its complexity, identify exploitable structures and develop new algorithms – from mathematically based heuristics to learning-based approaches (e.g. Reinforcement Learning). For prototyping and simulations, use Pythonund performancecritical components to implement in C/C++. In addition, you will examine whether and when trained policies (e.g. via Reinforcement Learning) surpass static heuristics and under what conditions these are executable in real time on embedded hardware. You analyze the interaction between scheduling decisions and security mechanisms (e.g. memory insulation). You implement your methods as part of a runtime system on embedded hardware, for example RISC-V-SoC, ARM Cortex-A/M, FPGA-based systems and evaluate them with real AI workloads (transformer reference). You collect quantitative measurement data on latency, throughput and energy efficiency. In regular project meetings and research stays in Taiwan, you agree with hardware teams and integrate your runtime system into the overall architecture. You publish at international conferences (e.g. MLSys, ASPLOS, RTAS, NeurIPS Systems Track, DATE) and in professional journals. You will supervise Bachelor/Master's and Student Assistants on part aspects of your topic.

Minimum qualification: Very well completed scientific studies (Master/Uni Diploma) in electrical engineering, computer science, technical computer science, physics, mathematics or comparable C/C++ and Python programming skills Experience with machine learning, ideally transformer architectures Structured working to get involved in new scientific topics and enjoy collaborative work in an interdisciplinary and international team that lives on fresh ideas Very good knowledge of German and English for cooperation in an international environment Required qualifications: Experience with Reinforcement Learning and/or Numerical Optimization Knowledge of AI frameworks (PyTorch, Jax, ONNX, Huggingface Transformers etc.) First experience with embedded software development

Doctorate within 3 years in an application-oriented topic at the interface of algorithmic optimization, AI systems and edge computing. You can access institute-owned electronics laboratories, computing infrastructure and EDA toolchains for your practical work. Teaching obligations will not be transferred to you. The doctoral degree is awarded by the University of Duisburg-Essen As part of the international STICAM project, you will participate in multi-week research stays at the Taiwanese partner institutions and publish them at international conferences. In addition to the support provided by an institute-internal specialist, a regular organized exchange takes place on the scientific state of your work within the framework of a doctorate/paternal body. During the promotion period, you will be supported by accompanying offers. For example, you can benefit from regular doctoral student coaching and learn to safely apply professional methods of project management during your doctorate and to support the project acquisition. If the dissertation is submitted in due time, the option is to further deepen the research work within the framework of a connection agreement or to switch to other areas. The full-time post as a doctoral student with half-remuneration provides 50% of the time for your doctoral degree and 50% for cooperation in research projects as a scientific assistant. Flexible working hours (running time with integrated core working time from 9:30 to 15:00, Friday from 9:30 to 13:00) and mobile work on up to two days a week for better reconciliation of work and private life Occupational pension (VBL) and grant to the Deutschland-Ticket Job Very good transport with public transport/car and free parking and bicycle parking for employees Family and Occupational Compatibility Support: Mit-Kind-Office, Kindernotpflege und Beratungs zu Homecare-Eldercare etc. in cooperation with the pme family service Corporate Benefits: Benefits Offers from renowned manufacturers and brands