Adam Moody
Computer Scientist at Lawrence Livermore National Laboratory
Livermore, California, United States
Summary
🤩
Rockstar🎓
Top SchoolAdam Moody is a computer scientist with 18 years of experience building distributed, high-performance communication and data storage algorithms for Top500-class supercomputers at Lawrence Livermore National Laboratory. He leads multiple open-source projects used across supercomputer centers and contributes to prominent communities like Spack and DeepSpeed, focusing on portability, performance, and large-scale training workflows. His work spans low-level packaging and build compatibility to ML-focused optimizations—enabling CPU Adam, async I/O, and parallel checkpointing for DeepSpeed and scaling data preprocessing for Megatron-DeepSpeed with MPI. Based in Livermore, CA, he combines hands-on engineering with production-grade research impact, and unusually bridges package management, HPC I/O, and large-scale deep learning pipelines.
18 years of coding experience
Bachelor of Science (B.S.), Bachelor of Science (B.S.) at The Ohio State University
Github Skills (38)
pytorch10
c-language10
python10
package-management10
data-parallel10
mpi10
machine-learning10
datasets10
data-preprocessing10
package-manager-tool10
deeplearning-ai10
data-parallelism10
deep-learning10
gpu10
huggingface-datasets10
Programming languages (11)
Github contributions (5)
bigscience-workshop/Megatron-DeepSpeed
Aug 2021 - Sep 2021
Ongoing research training transformer language models at scale, including: BERT & GPT-2
Role in this project:
ML Engineer
ML EngineerContributions:181 reviews, 7 commits, 10 PRs in 1 month
Contributions summary:Adam primarily contributed to the data preprocessing pipeline for training transformer language models. Their work focused on integrating Hugging Face Datasets, optimizing data loading using NumPy, and parallelizing the preprocessing steps with MPI. They also added features for shuffling and splitting datasets, and enhanced the data merging process for the indexed dataset files. These improvements aimed to improve the efficiency and scalability of the data preparation stage for large-scale language model training.
nlptransformersbertongoingscale
spack/spack
Jul 2014 - Feb 2022
A flexible package manager that supports multiple versions, configurations, platforms, and compilers.
Role in this project:Backend Developer
Backend DeveloperContributions:14 reviews, 41 commits, 31 PRs in 7 years 8 months
Contributions summary:Adam primarily contributed to the `spack/spack` repository by adding and updating package definitions. This included adding new packages like `gperftools`, `protobuf`, `glog`, `gflags`, `snappy`, `leveldb`, and `lmdb`, as well as updating existing packages such as `boost` and `scr`. The commits reflect a focus on expanding the available software packages managed by spack and maintaining its compatibility with other tools.
compilerspythonradiussplatformslinux
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.