Tristan Webb PhD (He/him)

Dr. Tristan J. Webb is a Principal Machine Learning Research Engineer at d-Matrix whose 15-year career has been dedicated to re-engineering the foundational layers of AI for maximum efficiency and performance. With research cited over 780 times, his work operates at the critical intersection of deep learning theory and high-performance, hardware-aware systems. He is a leading expert in model optimization, specializing in the compiler, numerical format, and algorithmic solutions required to scale generative AI.

His reputation for foundational systems work was established at Intel's Artificial Intelligence Products Group. There, Dr. Webb co-authored "Intel nGraph," a flexible intermediate representation and compiler designed to execute deep learning models efficiently across diverse hardware backends. He was also a key co-author of the highly-cited NeurIPS 2017 paper "Flexpoint," which introduced a novel, adaptive numerical format for efficient model training, directly addressing a core challenge in hardware-aware ML.

At d-Matrix, Dr. Webb now leads research targeting the primary bottleneck in modern AI: efficient generative AI inference. His current work is defining scaling laws for Large Language Model (LLM) quantization, research in efficient reinforcement learning with verifiable rewards, and generating new intellectual property in hardware-aware system optimization. Dr. Webb is a leading contributor to dmx-compressor (https://github.com/d-matrix-ai/dmx-compressor), a Pytorch framework aimed at deep neural net co-design for custom hardware accelerators.

Dr. Webb's unique perspective is built on a diverse technical foundation, beginning with a PhD in Complexity Science focused on computational neuroscience, followed by hands-on systems programming experience as a Haskell Application Engineer. During this time, he gained experience in software consulting, Haskell and Android development, acquired a passion for Kubernetes and DevOps, and engaged in exciting quantitative projects.

Tristan also served as a Postdoctoral Scientist at UC San Diego from September 2013 to October 2014. While at UCSD, Tristan conducted research in the field of Dynamical System in Neuroscience and conducted engineering work on the Neuromorphic Chip, NeuroDyn, previously developed at UCSD.

Earlier in his career, Tristan worked at University of Warwick as a Research Assistant from September 2012 to July 2013 in the field of Computation Vision. Prior to that, he was a PhD Researcher at the University of Warwick Complexity Science DTC from November 2008 to September 2012, where Tristan completed his thesis "Biologically Plausible Attractor Networks" and earned his PhD in Complexity Science. Before embarking on his PhD, Tristan earned his Master's in Complexity Science at the University of Warwick Complexity Science DTC from August 2007 to July 2008.

Additionally, during his time in graduate school, Tristan has experience as a Distance Education Instructor at Maharishi University of Management from May 2008 to July 2009 and as an Intern at Hewlett-Packard Laboratories from August 2008 to November 2008.

Tristan Webb, PhD, has a solid education history in the field of Complexity Science and Computer Science. In 2003, he enrolled at Maharishi International University and pursued his Bachelor of Science (B.Sc.) degree in Mathematical Sciences, which he successfully completed in 2006. Following this, Tristan continued his academic journey at the same institution, earning a Master of Science (MS) degree in Computer Science from 2005 to 2007, studying Parallel Programming, Operating Systems, and Software Engineering.

In his spare time, Tristan enjoys engaging in traditional darkroom photography, playing many classic games.