Marc Van Zee
Copenhagen, Capital Region of Denmark
Summary
🤩
Rockstar🎓
Top SchoolMarc Van Zee is a machine learning engineer and researcher with 11 years of experience focused on NLP, compositional generalization, and deep learning, currently based in Copenhagen. He leads development of Flax, a flexible neural network library for JAX used widely in ML research, and has contributed performance-focused changes to the JAX/XLA stack. Marc is the author of the CFQ dataset for measuring compositional generalization and has integrated it into T5 training and evaluation workflows at Google Research. With a PhD in Computer Science and a top-ranked AI master's, he blends rigorous academic training with hands-on engineering—often tackling low-level compiler interactions and sequence-to-sequence model implementations that bridge research and production.
10 years of coding experience
Doctor of Philosophy (PhD), Computer Science, 5/5 (outstanding), Doctor of Philosophy (PhD), Computer Science, 5/5 (outstanding) at University of LuxembourgPropedeuse, Computer Science, Propedeuse, Computer Science at Eindhoven University of TechnologyMaster, Artificial Intelligence, 9.3/10 (cum laude), Master, Artificial Intelligence, 9.3/10 (cum laude) at Universiteit Utrecht
Luxembourgish, German, Dutch, English
Stackoverflow
1,405reputation
124kreached
3answers
14questions
Github Skills (23)
python10
machine-learning10
flax10
data-preprocessing10
sequence-modeling10
lstm10
deep-learning10
tensorflow10
xla10
nlp10
jax10
natural-language-processing9
compiler-optimization9
distributed-computing8
linear-algebra8
Programming languages (6)
Github contributions (5)
google/flax
Mar 2020 - Jan 2023
Flax is a neural network library for JAX that is designed for flexibility.
Role in this project:
ML Engineer
ML EngineerContributions:6 releases, 454 reviews, 408 commits in 2 years 11 months
Contributions summary:Marc contributed to an LSTM encoder/decoder architecture on a sequence-to-sequence addition task within the context of the Flax library, suggesting involvement in machine learning development. The primary focus was on implementing and testing an LSTM-based sequence-to-sequence model, as evidenced by the code changes and the inclusion of character encoding and decoding classes. The contributions involved developing, testing, and debugging the model within the Flax framework, highlighting their understanding of deep learning and sequence modeling.
deep-learningneural-networksneural-networkflaxjax
jax-ml/jax
Nov 2020 - Jan 2023
Composable transformations of Python+NumPy programs: differentiate, vectorize, JIT to GPU/TPU, and more
Role in this project:Back-end Developer
Back-end DeveloperContributions:42 reviews, 58 commits, 13 PRs in 2 years 2 months
Contributions summary:Marc primarily contributed to the JAX library, focusing on improvements and updates related to the XLA (XLA - Accelerated Linear Algebra) compiler for machine learning. Their commits include internal changes, updates to comments, and modifications to the code related to XLA's interaction with the TensorFlow (TF) framework. The changes appear to involve optimizing the handling of different data types and padding within the XLA compiler. These contributions suggest a focus on enhancing the performance and capabilities of the JAX library, particularly in areas like distributed computing and support for machine learning tasks.
pytorchpythonjitautomatic-differentiationgpu
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.