Martin Maas

Staff Research Scientist, Google DeepMind

1600 Amphitheatre Pkwy, Mountain View, CA

E-Mail: <firstname>@martin-maas.com

I am a Staff Research Scientist at Google DeepMind. My primary research interests are in managed language runtime systems, operating systems and computer architecture. I am interested in the entire stack from the hardware to the programming systems layer. My current focus area is how to leverage machine learning to improve computer systems. An overview of my approach to this area can be found here.

Before joining Google, I completed my PhD in the Electrical Engineering and Computer Sciences department at UC Berkeley, working with Krste Asanović and John Kubiatowicz. My PhD research focused on warehouse-scale computers. I worked and collaborated across areas and built real systems that involve large system-level codebases as well as hardware-level RTL. I have applied this approach to domains ranging from security to managed languages. During my PhD, I built a secure processor that provides memory-trace obliviousness (a new security property) and can be targeted by a custom compiler, a distributed language runtime system that coordinates JVMs on different nodes in a cluster, and worked on hardware support for garbage collection. I have also built research infrastructure, including FPGA implementations of hardware based on the RISC-V ISA.

Before coming to UC Berkeley, I completed my undergraduate degree at the University of Cambridge. In my undergraduate research, I investigated the challenges and bottlenecks of implementing a Java Virtual Machine for the Barrelfish Operating System. I was supervised by Ross McIlroy and Tim Harris from Microsoft Research, Cambridge.

During my time in high-school, I was an active participant in science and programming competitions. I was on the German team for the International Olympiad of Informatics (IOI) and represented Germany at the International Science and Engineering Fair (ISEF).

Selected Publications

LAVA: Lifetime-Aware VM Allocation with Learned Distributions and Adaptation to Mispredictions, Jianheng Ling, Pratik Worah, Yawen Wang, Yunchuan Kong, Chunlei Wang, Clifford Stein, Diwakar Gupta, Jason Behmer, Logan A. Bush, Prakash Ramanan, Rajesh Kumar, Thomas Chestna, Yajing Liu, Ying Liu, Ye Zhao, Kathryn S. McKinley, Meeyoung Park, Martin Maas, 2025, Eighth Annual Conference on Machine Learning and Systems (MLSys'25), Santa Clara, California (to appear) Preprint
A Bring-Your-Own-Model Approach for ML-Driven Storage Placement in Warehouse-Scale Computers, Chenxi Yang, Yan Li, Martin Maas, Mustafa Uysal, Ubaid Ullah Hafeez, Arif Merchant, Richard McDougall, 2025, Eighth Annual Conference on Machine Learning and Systems (MLSys'25), Santa Clara, California (to appear) Preprint
TelaMalloc: Efficient On-Chip Memory Allocation for Production Machine Learning Accelerators, Martin Maas, Ulysse Beaugnon, Arun Chauhan, Berkin Ilbeyi, International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS '23), Vancouver, Canada, March 2023 Paper
Distilling the Real Cost of Production Garbage Collectors, Zixian Cai, Steve Blackburn, Mike Bond, Martin Maas, 2022 IEEE International Symposium on Performance Analysis of Systems and Software (ISPASS '22), May 2022 Paper
Adaptive Huge-Page Subrelease for Non-moving Memory Allocators in Warehouse-Scale Computers, Martin Maas, Chris Kennelly, Khanh Nguyen, Darryl Gove, Kathryn S. McKinley, Paul Turner, International Symposium on Memory Management (ISMM '21), June 2021 Paper
Learning on Distributed Traces for Data Center Storage Systems, Giulio Zhou, Martin Maas, Conference on Machine Learning and Systems 2021 (MLSys '21), April 2021 Paper
A Taxonomy of ML for Systems Problems, Martin Maas, IEEE Micro, Sept/Oct 2020 Paper
Learning-based Memory Allocation for C++ Server Workloads, Martin Maas, David G. Andersen, Michael Isard, Mohammad Mahdi Javanmard, Kathryn S. McKinley, Colin Raffel, International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS '20), Lausanne, Switzerland, March 2020 Paper | Talk Video (SIGPLAN Research Highlight, CACM Research Highlight)
A Hardware Accelerator for Tracing Garbage Collection, Martin Maas, Krste Asanović, John Kubiatowicz, 45th International Symposium on Computer Architecture (ISCA'18), Los Angeles, California, June 2018 Paper (Selected as one of IEEE Micro's Top Picks from the 2018 Computer Architecture Conferences)
Taurus: A Holistic Language Runtime System for Coordinating Distributed Managed-Language Applications, Martin Maas, Krste Asanović, Tim Harris, John Kubiatowicz, International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS '16), Atlanta, Georgia, April 2016 Paper
GhostRider: A Hardware-Software System for Memory Trace Oblivious Computation, Chang Liu, Austin Harris, Martin Maas, Michael Hicks, Mohit Tiwari, Elaine Shi, International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS '15), Istanbul, Turkey, March 2015 Paper (Winner of the Best Paper Award)
PHANTOM: Practical Oblivious Computation in a Secure Processor, Martin Maas, Eric Love, Emil Stefanov, Mohit Tiwari, Elaine Shi, Krste Asanović, John Kubiatowicz, Dawn Song, ACM Conference on Computer and Communications Security (CCS '13), Berlin, Germany, November 2013 Paper (Finalist for NYU-Poly (formerly AT&T) Best Applied Security Paper Award 2013)
GPUs as an Opportunity for Offloading Garbage Collection, Martin Maas, Philip Reames, Jeffrey Morlan, Krste Asanović, Anthony D. Joseph, John Kubiatowicz, International Symposium on Memory Management (ISMM '12), Beijing, China, June 2012 Paper

Google Sites

Report abuse