Dettmers Receives Google ML and Systems Junior Faculty Award
Adam KohlhaasFriday, August 1, 2025Print this page.
Tim Dettmers, an assistant professor in Carnegie Mellon University's machine learning and computer science departments, has been named a recipient of the inaugural Google ML and Systems Junior Faculty Award. The award recognizes early career faculty whose work advances scalable and efficient machine learning systems.
Dettmers is among more than 50 assistant professors across 27 U.S. universities selected for the award, which includes $100,000 in unrestricted funding to support research at the intersection of machine learning and systems. Recipients were chosen by a panel of Google engineers and researchers for their contributions to trustworthy, high-performance computing across the technology stack — from algorithms to hardware infrastructure.
At CMU's School of Computer Science, Dettmers' research explores sparsity in deep learning, with particular focus on sparse quantization and training techniques that aim to make large-scale AI models significantly more efficient and accessible. His work investigates how models can be compressed and selectively activated during inference and training, allowing for faster, more cost-effective deployment without compromising performance.
In sparse quantization, Dettmers develops methods that reduce the memory footprint of AI models by encoding parameters in lower precision formats — sometimes as little as four bits per element — while strategically retaining higher precision for critical components. His sparse training research centers on mixture-of-experts architectures, which route computation through only a small, relevant subset of the model depending on the input. These techniques have the potential to cut computational costs by an order of magnitude, making it feasible to train and deploy complex transformer models at scale.
The award highlights growing industry investment in academic research, especially as AI systems continue to push the limits of computational infrastructure. For researchers like Dettmers, early career funding plays a critical role in advancing foundational work that underpins the next generation of machine learning systems.
For more information, visit the Google Blog.
Aaron Aupperlee | 412-268-9068 | aaupperlee@cmu.edu