Dirk Groeneveld
Member Of Technical Staff at Off Camber Data
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolDirk Groeneveld is a seasoned software engineer and technical leader with 12+ years building scalable backend systems, now a Member of Technical Staff at Microsoft AI and founder of a motorsport data startup focused on real-time acquisition and analysis. He spent many years at the Allen Institute for AI advancing NLP tooling and model training infrastructure—contributions include gradient accumulation, transformer tokenization fixes, and T5 model improvements—and has practical ML engineering experience reflected in test and QA work on the popular EleutherAI lm-evaluation-harness. Dirk pairs hands-on engineering with DevOps discipline, having modernized build and release processes for projects like pdffigures2 and resolved complex dependency and publishing challenges. His background spans startups and large tech (Microsoft, Amazon) and a track record of shipping production-ready systems that bridge research and product. An entrepreneur at heart, he brings both leadership and deep technical craft to teams building data-intensive and ML-driven products.
11 years of coding experience
18 years of employment as a software developer
University of California, Irvine
Stackoverflow
2,577reputation
122kreached
66answers
18questions
Github Skills (25)
dependency-management10
pytorch10
pytest10
python10
testing10
machine-learning10
sbt10
scala10
transformer-models10
release-management10
deep-learning10
natural-language-processing10
build-automation10
nlp10
language-model9
Programming languages (12)
Github contributions (5)
allenai/allennlp
Dec 2019 - Oct 2022
An open-source NLP research library, built on PyTorch.
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:14 releases, 495 reviews, 225 commits in 2 years 10 months
Contributions summary:Dirk primarily focused on improvements to the AllenNLP library related to training, model definition, and model performance. Contributions included implementing gradient accumulation features, addressing issues with tokenization and vocabulary handling for transformer-based models, and incorporating optimizers. The user also worked on refining and adding features to the T5 model implementation within the library.
nlppytorchtransformerspythonbert
allenai/pdffigures2
Jan 2017 - May 2019
Given a scholarly PDF, extract figures, tables, captions, and section titles.
Role in this project:Backend Developer & DevOps Engineer
Backend Developer & DevOps EngineerContributions:42 commits, 15 PRs, 15 pushes in 2 years 4 months
Contributions summary:Dirk primarily contributed to the project by modifying build configurations and dependency management, specifically related to SBT plugins and publishing settings. They addressed build-related issues by switching to new plugins and updating versions. Additionally, the user implemented release management procedures and configured repository access, demonstrating DevOps skills. Furthermore, they added filters and made optimizations to improve performance and overall project quality.
pdffiguresextract-pdfcaptionsscholarly
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.