Hailey Schoelkopf
Member Of Technical Staff at Anthropic
Greater Boston United States
Summary
🤩
Rockstar🎓
Top SchoolHailey Schoelkopf is a Member of Technical Staff and research-focused software engineer specializing in Large Language Models, with five years of experience spanning efficient at-scale training, model interpretability, evaluation, and deployment. She has driven open-source contributions and tooling for prominent projects like GPT-NeoX and EleutherAI’s Pythia hub—building sharding utilities, checkpoint converters to Hugging Face, and evaluation integrations that bridge research and production. At EleutherAI she published and managed collaborative research while coordinating volunteers and staff; she now applies that blend of hands-on MLOps and scientific rigor at Anthropic. Trained in mathematics and computer science at Yale, she pairs strong theoretical grounding with practical systems work, and often focuses on the underappreciated plumbing—data handling and checkpointing—that makes large-model research reproducible and scalable.
5 years of coding experience
2 years of employment as a software developer
Bachelor's degree, Mathematics and Computer Science, Bachelor's degree, Mathematics and Computer Science at Yale University
Github Skills (16)
transformers10
data-handling10
machine-learning10
pytorch10
python10
scripting10
numpy10
checkpointing9
argument-parsing9
gpt9
command-line-interface9
checkpoint9
language-model9
huggingface8
huggingface-hub8
Programming languages (3)
Github contributions (5)
EleutherAI/pythia
Oct 2022 - Jan 2023
The hub for EleutherAI's work on interpretability and learning dynamics
Role in this project:
MLOps Engineer
MLOps EngineerContributions:24 reviews, 136 commits, 57 PRs in 3 months
Contributions summary:Hailey primarily contributed to the development of scripts for managing and manipulating large model data files within the context of a machine learning project. They implemented tools for sharding and un-sharding model data (.bin files), crucial for handling large datasets often encountered in machine learning. Furthermore, the user created a batch viewer script, indicating work on data inspection and analysis within the training pipeline. Their work emphasizes data handling, a key aspect of model training and deployment, suggesting an MLOps focus.
EleutherAI/gpt-neox
Jul 2022 - Jan 2023
An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries
Role in this project:ML Engineer
ML EngineerContributions:25 reviews, 194 commits, 48 PRs in 6 months
Contributions summary:Hailey made several contributions focused on improving the evaluation and deployment of models within the GPT-NeoX framework. This included adding logging capabilities using wandb, integrating wildcard pattern matching for task selection in the evaluation harness, and developing a script to convert NeoX checkpoints to Hugging Face (HF) model format. The user also addressed bugs and made code refinements related to model conversion and checkpointing behavior, ensuring correct operation.
pytorchlanguage-modeltransformersdeepspeedmodel-parallel
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.