Stella Biderman
Executive Director at EleutherAI
Alexandria, Virginia, United States
Summary
🤩
Rockstar🎓
Top SchoolStella Biderman is an executive director and researcher with a decade of experience blending mathematics, theoretical computer science, and applied AI to advance large language models and social network analysis. She led EleutherAI’s transition from a volunteer collective to a formal research nonprofit and helped train and open-source landmark models such as GPT-Neo, GPT-NeoX-20B, and BLOOM. With a background spanning Booz Allen Hamilton and academic research at the University of Chicago, she applies rigorous theory to practical problems in malware detection, sociotechnical analysis, and ethical implications of computational decision-making. An active open-source contributor, she has improved core training pipelines and evaluation frameworks used by major language-model projects, reflecting deep hands-on expertise in model-parallel training and DevOps for large-scale transformers. Her combined MS in Computer Science and dual BS degrees in Mathematics and Philosophy underpin a rare mix of technical depth and conceptual thinking about AI governance and democratization.
10 years of coding experience
7 years of employment as a software developer
Master of Science - MS, Computer Science, Master of Science - MS, Computer Science at Georgia Institute of TechnologyBachelor of Science - BS, Mathematics, Bachelor of Science - BS, Mathematics at University of Chicago
Stackoverflow
141reputation
4kreached
4answers
2questions
Github Skills (24)
transformers10
language-model10
distributed-training10
python10
scripting10
script10
deepspeed10
natural-language-processing10
sh10
transformer10
evaluation-framework10
gpt10
nlp10
shell10
machine-learning9
Programming languages (11)
Github contributions (5)
EleutherAI/gpt-neox
Dec 2020 - Jan 2023
An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries
Role in this project:
Back-end & DevOps Engineer
Back-end & DevOps EngineerContributions:3 releases, 278 reviews, 743 commits in 2 years 1 month
Contributions summary:Stella's contributions primarily revolve around improving and refactoring the training pipeline for the GPT-NeoX model, especially with pipeline parallelism. They made numerous changes to the core training script, configuration files, and shell scripts used for training, optimizing DeepSpeed configurations to better utilize ZeRO stages and continuous gradients. Furthermore, the user was also involved in setting up the environment for the project. The user also modified the bash scripts for deployment.
pytorchlanguage-modeltransformersdeepspeedmodel-parallel
EleutherAI/gpt-neo
Aug 2020 - Feb 2022
An implementation of model parallel GPT-2 and GPT-3-style models using the mesh-tensorflow library.
Role in this project:ML Engineer
ML EngineerContributions:1 release, 30 reviews, 101 commits in 1 year 6 months
Contributions summary:Stella made several updates to the `create_tfrecords.py` script, which is likely involved in preparing data for training or inference with the GPT-Neo models. These updates include adding a clarifying comment and fixing a parenthesis issue, suggesting the user is working on the data processing pipeline. Additionally, the user fixed a notebook file, indicating they were also contributing to the project's documentation or example code. The user also performed a merge.
meshlanguage-modeltransformersmodel-parallelparallel
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.