Milo Cress
Member Of Technical Staff at Anthropic
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolMilo Cress is a software engineer with nine years of experience building data-heavy ML infrastructure and production training pipelines, currently working on pretraining at Anthropic. He has a strong background in streaming dataset engineering and LLM finetuning workflows from contributions to MosaicML projects and Databricks’ LLM Foundry, improving robustness, tokenization, and chat-formatted finetuning data. A former co‑founder and research engineer, Milo pairs startup grit with research-grade rigor, having optimized multi-rank streaming dataloaders and introduced practical notebooks for synthetic face datasets. Based in New York and trained in mathematics at MIT, he combines a curiosity for taking things apart to understand systems with hands-on experience across research, quant engineering, and product-focused roles.
9 years of coding experience
3 years of employment as a software developer
Bachelor's degree, Mathematics, Bachelor's degree, Mathematics at Massachusetts Institute of Technology
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (22)
pytorch10
python10
machine-learning10
ml10
llm10
deep-learning10
natural-language-processing10
data-processing10
neural-network10
nlp10
data-engineering9
tokenize9
pytest9
tokenizer9
computer-vision9
Programming languages (5)
Github contributions (5)
mosaicml/llm-foundry
Jan 2024 - Mar 2025
LLM training code for Databricks foundation models
Role in this project:
Back-end Developer
Back-end DeveloperContributions:2 releases, 106 reviews, 88 PRs in 1 year 2 months
Contributions summary:Milo primarily contributed to the data processing and model training aspects of the LLM foundry, as evident by the numerous changes to the finetuning data loading, and tokenization processes. The user added support for chat formatted finetuning data, which involved modifying existing code, adding new functionalities, and refactoring test cases. Additionally, the user focused on improving the robustness of the codebase.
databricksfoundation-modelsllmdeep-learningneural-networks
mosaicml/composer
Jun 2022 - Jul 2022
Supercharge Your Model Training
Role in this project:ML Engineer
ML EngineerContributions:1 release, 55 reviews, 7 commits in 29 days
Contributions summary:Milo primarily contributes to the project by adding and modifying code related to streaming datasets, particularly for machine learning training pipelines. They introduced a notebook demonstrating a streaming dataloader for the FaceSynthetics dataset, enhancing data loading efficiency. Further contributions include improvements to the streaming dataset functionality by adding compression options and fixing issues related to multi-rank environments. The user also made modifications to existing dataset configurations and related testing procedures.
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.