David Wadden is a research scientist and NLP engineer with 11 years of experience building data-driven systems for information extraction and coreference. He contributes to high-impact open-source projects such as AllenNLP and dygiepp, implementing robust dataset loaders, numerical-stability fixes, and batching/pruning improvements for variable-length inputs. His work sits at the intersection of model engineering and preprocessing infrastructure, ensuring that complex span-based NER, relation, and event extraction models run reliably in practice. Based in the United States, he combines deep research sensibilities with pragmatic engineering—often tackling subtle edge cases (e.g., NaNs and division-by-zero) that keep production research pipelines healthy.
An open-source NLP research library, built on PyTorch.
Role in this project:
ML Engineer
Contributions:5 commits, 7 PRs, 13 comments in 1 year 5 months
Contributions summary:David primarily contributed to the AllenNLP library, focusing on machine learning model-related features and bug fixes. They implemented enhancements to the `Pruner` class, enabling it to handle variable-length inputs within a minibatch, addressing a specific need in NLP tasks such as coreference resolution. Moreover, the user addressed potential numerical instability issues related to the `MismatchedEmbedder` by correcting a division-by-zero error and adding fixes for NaNs. The user also worked on the Tensorboard writer to address logging edge cases.
Contributions:36 commits, 22 pushes, 1 branch in 1 year 2 months
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.
Request Free Trial
David Wadden - Research Scientist at University of Washington