Philip May
Senior Data Scientist
Brunswick, Lower Saxony, Germany
Summary
🤩
RockstarPhilip May is a Senior Data Scientist with 10 years of experience driving NLP and ML engineering work inside large enterprises, currently at Deutsche Telekom after a long tenure leading AI & Data Analytics at T-Systems. He is a prolific open-source contributor with impactful changes to flagship projects like Hugging Face Transformers, Sentence-Transformers, Optuna and ELECTRA—improving tokenization, few-shot training, hyperparameter checks, and pretraining pipelines. Philip blends research-aware model optimization (AMP, embedding normalization, loss choices) with production-minded robustness (test automation, refactors, memory fixes), making models both performant and maintainable. Based in Brunswick, Germany, he combines deep NLP expertise with DevOps and backend skills, and often opts for practical tooling improvements—like flexible tokenizer options and evaluation callbacks—that quietly lift developer productivity.
10 years of coding experience
17 years of employment as a software developer
Github Skills (40)
dependency-management10
tokenize10
sentencepiece10
pytorch10
language-model10
pytest10
sentence-transformers10
py10
python10
package-management10
setuptools10
testing10
command-line-arguments10
machine-learning10
data-preprocessing10
Programming languages (22)
Github contributions (5)
deepset-ai/FARM
Mar 2020 - Sep 2020
:house_with_garden: Fast & easy transfer learning for NLP. Harvesting language models for the industry. Focus on Question Answering.
Role in this project:
ML Engineer
ML EngineerContributions:9 reviews, 34 commits, 25 PRs in 6 months
Contributions summary:Philip contributed significantly to the `farm` repository, which focuses on transfer learning for NLP, specifically in question answering. Their contributions involved adding and fixing code related to language model fine-tuning examples, improving documentation with a focus on data loading, and addressing issues related to early stopping, training progress bar display, and model loading. Further contributions included options to specify text column names and using fast HF tokenizer. Their work appears primarily centered around improving the framework's functionality and usability for NLP tasks.
bertxlnet-pytorchnatural-language-processingtransformersharvesting
optuna/optuna
Feb 2020 - Jul 2020
A hyperparameter optimization framework
Role in this project:Back-end Developer & Test Automation Engineer
Back-end Developer & Test Automation EngineerContributions:1 review, 116 commits, 11 PRs in 4 months
Contributions summary:Philip primarily contributed to the development and testing of the `optuna` framework. Their work focused on implementing new features related to distribution checks and refining existing functionalities, specifically within the `trial.py` and `distributions.py` files. The user also created and expanded tests within `tests/test_trial.py` to ensure the reliability of the distribution-related functionalities. They addressed issues with code formatting and type hinting, and improved the overall robustness of the framework.
pythonoptimization-frameworkparallelhyperparameteroptimization
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.