Dean Wyatte
Principal Machine Learning Engineer (NLP) at Square
Boulder, Colorado, United States
Summary
👤
Senior🎓
Top SchoolDean Wyatte is a Principal Machine Learning Engineer specializing in NLP with a decade of experience building production-scale ML systems from research prototypes to customer-facing services. Based in Boulder, he leads technical direction for customer support ML at Square, drawing on a strong academic foundation (Ph.D. in Cognitive Neuroscience) and a history of published research that informs biologically inspired models and human-centric solutions. His background spans deep learning for image recognition, large-scale similarity search, real-time systems, and neuromorphic computing, with practical deployments in gaming, security, and enterprise platforms. An active contributor to Hugging Face Transformers, he has improved token masking utilities and ONNX export robustness, highlighting attention to interoperability and production export constraints. He combines rigorous experimental methods from neuroscience with engineering pragmatism to deliver scalable, interpretable models that account for human behavior. Colleagues rely on him for bridging cutting-edge research and reliable, high-throughput ML infrastructure.
10 years of coding experience
19 years of employment as a software developer
Bachelor of Science (B.S.) with Honors Computer Science Cognitive Science, Bachelor of Science (B.S.) with Honors Computer Science Cognitive Science at Indiana University BloomingtonDoctor of Philosophy (Ph.D.) Cognitive Neuroscience, Doctor of Philosophy (Ph.D.) Cognitive Neuroscience at University of Colorado Boulder
Github Skills (13)
pytorch10
machine-learning10
transformer10
nlp10
tensorflow10
onnx10
python10
exporter9
data-export9
exports9
testing8
jax6
flax6
Programming languages (9)
Github contributions (5)
huggingface/transformers
Nov 2021 - Oct 2022
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
Role in this project:
ML Engineer
ML EngineerContributions:2 reviews, 2 commits, 6 PRs in 11 months
Contributions summary:Dean primarily contributed to the `DataCollatorForWholeWordMask` class, enhancing its functionality and ensuring compatibility with different tensor types ("np", "tf"). They also addressed ONNX export issues related to dynamic input shapes for various models, including causal LM sequence classifiers, and optimized ONNX export with respect to sequence length. Further contributions include updating documentation and making tests more robust within this machine-learning focused transformer library.
pythonbertspeech-recognitionstate-of-the-artflax
dwyatte/dotfiles
Feb 2016 - Feb 2025
Contributions:54 pushes, 1 branch in 9 years 1 month
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.