Curriculum Vitae
Tim Langer is a Research Associate at the Chair of Highly-Parallel VLSI Systems & Neuro-Microelectronics at TU Dresden. His research focuses on neuromorphic computing and efficient hardware architectures for machine learning, with a particular interest in mapping and optimizing neural networks on parallel and embedded systems. In addition to his research activities, he contributes to teaching in the areas of neural network hardware and neuromorphic systems.
Previously, he worked at Fraunhofer IIS/EAS and Infineon Technologies, where he gained extensive experience in embedded machine learning, signal processing, and hardware/software co-design for industrial and automotive applications. His work spans from algorithm development and optimization to low-level implementation on embedded and accelerator-based platforms.
Tim studied Electrical Engineering at TU Dresden, specializing in Information Technology. His academic background combines machine learning, signal processing, and digital hardware design, forming the foundation for his interdisciplinary work at the intersection of AI and hardware.
Projekte/Kooperationen innerhalb von CeTI, an denen Sie beteiligt sind:
- efficient surgical video processing on SpiNNaker2
What are the main topics or questions that drive your research?
I am interested in making AI algorithms more efficient to reduce their energy consumption. My particular interest is to increase inference efficiency for surgical video processing, considering aspects of event-based and neuromorphic computing. One key question is also how to implement these algorithms on neuromorphic hardware like SpiNNaker2.
What inspired you to pursue your current field of work?
During my studies, internships and other positions, I came in touch with AI on various levels reaching from digital hardware design for custom AI accelerators via low-level coding AI algorithms up to developing and training AI algorithms as well as creating application-specific datasets.
The most fascinating aspect for me was making AI algorithms run efficiently on custom hardware, as it requires technical creativity from algorithm design down to low-level programming. With the application of some tricks, you can achieve a significant performance boost in latency and energy efficiency. This can be very rewarding and helps to reduce the huge energy budget which AI algorithms currently require, resulting in a positive ecological impact.
On the application side, I am very interested in surgical video processing, as I think there is a lot of potential to reduce energy consumption and processing latency due to spatio-temporal correlation.
What excites you most about being part of CeTI?
I am excited for interdisciplinary exchange and to collaborate with great scientists within CeTI. As someone who grew up in the region, I am also looking forward to contributing to CeTI as a part of the current excellence period of TU Dresden.
Which challenge or question has recently sparked your curiosity?
One recent research question is how concepts of existing compression algorithms can be exploited for making AI algorithms for surgival video processing more efficient. Another research questions is how algorithms based on Mixture-of-Experts (MoE) can be optimized for neuromorphic hardware like SpiNNaker2.
How do you like to recharge or spend your time outside of work?
As a recreational activity, I play lead guitar in a rock band. Additionally, I like to swim and I am active in water rescue. I like to go outdoors for hiking and for photography.
