Michael Banfield
Senior Software Engineer at Google
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolMichael Banfield is a Senior Software Engineer in Seattle with 12 years of experience building scalable cloud and ML infrastructure, currently contributing to Google Maps after several years on Cloud TPU. He has deep hands-on expertise integrating TPUs with TensorFlow—working on model training, quantization, checkpointing, and robust preemption/error handling for large-scale TPU workloads. Prior roles at AWS and startups round out his cloud systems and software engineering pedigree, and his dual degrees in Software Engineering and Finance reflect a practical, data-informed approach to problem solving. Colleagues rely on him to bridge research-model work and production-ready tooling, particularly where performance and fault tolerance matter.
12 years of coding experience
8 years of employment as a software developer
Bachelor’s Degree, Commerce (Finance), Bachelor’s Degree, Commerce (Finance) at The University of Queensland
Github Skills (12)
quantization10
machine-learning10
error-handling10
tensorflow10
trainings10
python10
tpu10
modeling10
keras9
efficientnet8
resnet8
cicd3
Programming languages (7)
Github contributions (5)
tensorflow/tpu
Feb 2019 - Apr 2022
Reference models and tools for Cloud TPUs.
Role in this project:
ML Engineer
ML EngineerContributions:11 commits, 3 PRs, 5 pushes in 3 years 2 months
Contributions summary:Michael primarily contributed to model training and evaluation, specifically for the MnasNet, EfficientNet, and ResNet models, all within the context of TPUs. They made changes related to quantization during training, including adding fake quantization ops and enabling post-quantization. Furthermore, they were involved in integrating moving average variables and managing checkpoints for model export and initialization. The user also updated Keras colab notebooks.
cloud
tensorflow/estimator
May 2019 - Nov 2019
TensorFlow Estimator
Role in this project:ML Engineer
ML EngineerContributions:7 commits in 6 months
Contributions summary:Michael primarily focused on modifying the `tensorflow/estimator` repository related to TPU (Tensor Processing Unit) integration and error handling. Their contributions include adjustments to error handling mechanisms within the `ErrorRendezvous` class, specifically to ignore errors already handled by MonitoredSession. They also implemented and refined preemption hooks, enabling them when a TPU name is provided. Additionally, the user addressed issues related to cloud TPU preemption by enabling or disabling it based on whether the code is running on GCE.
deep-learningmachine-learningtensorflowtensorflow-estimatorestimator
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.