Gabriel Tseng
Machine Learning Engineer at allenai
Summary
🤩
RockstarGabriel Tseng is a Machine Learning Engineer with nine years of experience specializing in ML model interpretability and applications in energy and agriculture. He is an active open-source contributor to high-profile projects like PyTorch Captum and SHAP, where he improved testing infrastructure, implemented a PyTorch gradient explainer, added interim layer explanations, and resolved tricky memory and pooling bugs. Gabriel blends practical engineering—test automation and regression testing—with research-oriented work on attribution methods such as DeepLift and DeepExplainer. He favors robust, production-ready solutions that emphasize explainability and reproducibility across model architectures. Based in Canada, he brings domain-focused insight into deploying interpretable ML for real-world sectors like energy and agriculture. An attention to edge cases and memory/performance issues distinguishes his hands-on contributions.
8 years of coding experience
Github Skills (11)
unit-testing10
pytorch10
machine-learning10
explainable-artificial-intelligence10
interpretation10
deep-learning10
python10
testing9
mask-rcnn9
faster-rcnn9
feature-selection7
Programming languages (3)
Github contributions (5)
shap/shap
Aug 2018 - Nov 2019
A game theoretic approach to explain the output of any machine learning model.
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 35 commits, 15 PRs in 1 year 3 months
Contributions summary:Gabriel contributed significantly to the `shap/shap` repository, which focuses on explaining machine learning models. Their work involved implementing a PyTorch gradient explainer, including the necessary code and tests. The contributions extended to adding support for interim layer explanations within the PyTorch explainer, and enhancing the model with various layer types. Furthermore, the user added regression tests, and resolved memory issues within the PyTorch DeepExplainer.
explaininterpretabilityshapdeep-learningapproach
pytorch/captum
Sep 2019 - Oct 2019
Model interpretability and understanding for PyTorch
Role in this project:ML Engineer & QA Engineer/Test Automation Engineer
ML Engineer & QA Engineer/Test Automation EngineerContributions:11 commits, 3 PRs, 2 comments in 22 days
Contributions summary:Gabriel primarily contributed to improving the testing infrastructure and adding new tests for the `captum` library, focusing on the `DeepLift` algorithm. They added comprehensive tests for various model architectures like ConvNet, MaxPool1d, and MaxPool3d, ensuring accurate attribution calculations. Further, the user addressed a MaxPool1d error and implemented formatting changes to ensure code quality.
pytorchinterpretable-aifeature-importanceunderstandinginterpretability
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.