Davis Blalock
Research Scientist at Google DeepMind
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolDavis Blalock is a research scientist with an MIT PhD and 11 years of experience building high-performance machine learning systems, currently working on accelerator co-design and Gemini-class projects at Google DeepMind. He was an early engineer at MosaicML and later at Databricks, where he led model compression and helped ship DBRX, briefly the world’s best open-weight model. His work blends research and production: from fast algorithms for approximate matrix multiplication and time-series compression in academia to shipping LLM pretraining and inference tooling in industry. An active open-source contributor, he improves ML infrastructure and developer workflows—evident in contributions to MosaicML’s Composer and LLM-Foundry, plus a personal repo focused on 10x faster matrix/vector ops. Colleagues describe him as someone who consistently turns a year of research into an order-of-magnitude efficiency gain and who comfortably spans C++, Python, and ML systems. Based in San Francisco, he favors practical, automation-first fixes (Makefiles, pre-commit hooks, CI improvements) that boost reproducibility and developer velocity.
11 years of coding experience
12 years of employment as a software developer
MS + PhD Computer Science, MS + PhD Computer Science at Massachusetts Institute of TechnologyBachelor of Science (BS) Electrical Engineering; Cognitive Science, Bachelor of Science (BS) Electrical Engineering; Cognitive Science at University of Virginia
Github Skills (29)
pytorch10
github-ci10
gnu-make10
c-language10
integrate10
python10
model-driven10
machine-learning10
build-system10
makefile10
integrations10
cicd10
model-building10
automation10
deeplearning-ai10
Programming languages (3)
Github contributions (5)
dblalock/bolt
Feb 2017 - Jun 2022
10x faster matrix and vector operations
Role in this project:
Back-end Developer
Back-end DeveloperContributions:3 reviews, 268 commits, 4 PRs in 5 years 5 months
Contributions summary:Davis's commits primarily involve setting up and getting builds and profiling working under Bazel, including the implementation of a Python setup script. The code changes include additions to a C++ header file, suggesting the user is working with the core of the project's functionality, potentially involving matrix and vector operations. The commits demonstrate efforts to integrate build and installation processes, implying work within a development environment.
math10xcompressiondata-miningmachine-learning
mosaicml/composer
Oct 2021 - Oct 2022
Supercharge Your Model Training
Role in this project:ML Engineer
ML EngineerContributions:164 reviews, 34 commits, 52 PRs in 11 months
Contributions summary:Davis primarily contributed to the implementation and integration of machine learning models and related components within the MosaicML Composer framework. They added a new ResNet-9 model for CIFAR-10, including model definitions, configuration files, and integration with existing training pipelines. Their contributions involved modifying existing framework components, adding new model architectures, and ensuring compatibility with the testing framework, showcasing skills in deep learning model development and integration within a machine learning training library.
pytorchml-systemsdeep-learningneural-networksmachine-learning
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.