Matthew Douglas
ML Engineer at Hugging Face
Lebanon, Indiana, United States
Summary
🤩
Rockstar🎓
Top SchoolMatt is a Machine Learning Engineer based in Dublin with a decade of hands-on experience building and improving ML tooling and libraries. He contributes to high-profile open-source projects including Hugging Face, Keras, scikit-learn-contrib HDBSCAN, UMAP and transformers, where his work spans core algorithm tweaks, TensorFlow integration, and dataset-to-TF pipeline improvements. He has a track record of shipping pragmatic fixes—adding RaggedTensor support to Keras predict, introducing max_cluster_size to HDBSCAN, and enhancing to_tf_dataset for better padding and multi-label support—that improve real-world model usability. Comfortable in both Python and Cython, he focuses on robust testing, performance-aware changes, and edge-case handling such as sparse precomputed distances and variable-length inputs. Colleagues would describe him as the engineer who bridges research-grade algorithms and production-ready data pipelines within major ML ecosystems.
11 years of coding experience
14 years of employment as a software developer
BS Computer Science, BS Computer Science at Central Connecticut State University
Github Skills (32)
transformers10
sparse10
pytorch10
data-pipelines10
python10
imap10
scikit10
testing10
machine-learning10
hdbscan10
machine-learning-algorithms10
numpy10
datasets10
huggingface-transformers10
keras10
Programming languages (9)
Github contributions (5)
huggingface/transformers
Mar 2019 - Jan 2023
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
Role in this project:
ML Engineer
ML EngineerContributions:1635 reviews, 206 commits, 1096 PRs in 3 years 10 months
Contributions summary:Matthew's contributions center around the implementation of a new LM finetuning example, introducing and modifying files related to a pregenerated dataset for language model fine-tuning using the Hugging Face Transformers library. The code includes the definition of a custom dataset class, feature conversion, and training loop integration with the `BertForPreTraining` model. The user also addressed code style issues and improved test setup, indicating a focus on creating a functional and usable example.
audioinferencemachine-learning-modelsmultimodaltransformers
huggingface/notebooks
Sep 2021 - Jan 2023
Notebooks using the Hugging Face libraries 🤗
Role in this project:ML Engineer
ML EngineerContributions:19 reviews, 42 commits, 31 PRs in 1 year 4 months
Contributions summary:Matthew appears to be focused on developing and improving TensorFlow-based notebooks for training and utilizing Hugging Face libraries. Their contributions center on creating and updating language modeling notebooks, including those for causal language modeling, masked language modeling, and language modeling from scratch, all implemented with TensorFlow. The user also addressed mandatory data collator requirements and fixed typos, showcasing a focus on code functionality and usability within the context of Hugging Face's library ecosystem.
huggingface
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.