Anna Pfohl
Staff Software Engineer at MosaicML
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolAnna Pfohl is a Staff Software Engineer in San Francisco with eight years of experience building ML platforms that simplify training and serving large models. She blends hands-on ML engineering with platform design, contributing to open-source projects like MosaicML’s Composer and LLM-Foundry—adding robust MLflow logging, generation tooling, and async evaluation fixes that improve reproducibility and large-scale training workflows. At Databricks and MosaicML she’s focused on production-ready tooling for foundation models, bridging research code and reliable infrastructure. Her background in industrial engineering and an MEng in computer science gives her a systems-minded approach to scalability and observability. Colleagues know her for quietly fixing tricky edge cases (like single-tensor logging and async callbacks) that make model training reliable in real-world environments.
7 years of coding experience
6 years of employment as a software developer
Bachelor's degree Industrial Engineering, Bachelor's degree Industrial Engineering at Georgia Institute of TechnologyMaster of Engineering - MEng Computer Science, Master of Engineering - MEng Computer Science at Cornell Tech
Github Skills (12)
neural-network10
pytorch10
machine-learning10
mlflow10
mlops10
deep-learning10
python10
llm10
nlp9
tensorflow7
dockers7
docker7
Programming languages (4)
Github contributions (5)
mosaicml/llm-foundry
May 2023 - Oct 2024
LLM training code for Databricks foundation models
Role in this project:
ML Engineer
ML EngineerContributions:48 reviews, 43 PRs, 207 pushes in 1 year 5 months
Contributions summary:Anna contributed to the LLM training code for Databricks foundation models by implementing features and addressing issues related to the project's core functionality. They added MLFlow as a logger option, improving the project's ability to track and analyze model training runs. The user also worked on integrating and refining components related to model generation, including a script for bulk generation against an endpoint, streamlining the model evaluation and inference process. Their commits also included fixing an async eval callback, ensuring that async evaluation runs are launched reliably and efficiently.
deep-learningllmneural-networksnlppytorch
mosaicml/composer
Apr 2022 - Jun 2024
Supercharge Your Model Training
Role in this project:ML Engineer
ML EngineerContributions:33 reviews, 33 PRs, 58 pushes in 2 years 2 months
Contributions summary:Anna primarily contributed to the integration and enhancement of MLflow logging within the Composer framework. This included adding support for tags, ensuring correct experiment setup based on environment variables, and fixing how single value tensors are logged. They also worked on improving the Generate callback feature, adding logging and ensuring it is called at the end of training. Furthermore, the user contributed to upgrading the MCLI, a crucial element of this repository, incorporating improvements and bug fixes.
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.