Mike Dusenberry
Staff Research Engineer, Google DeepMind
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolMike Dusenberry is a staff research engineer at Google DeepMind with 13 years of experience building scalable Bayesian deep learning systems for high-stakes domains, especially medicine. His work bridges large neural nets, probabilistic modeling, and decision-making under uncertainty, and he helped develop Rank-1 Bayesian neural nets and Bayesian layers used in major ML libraries. A former M.D. student and co-author on COVID forecasting and clinical ML papers, he combines clinical insight with production-scale engineering across projects like TensorFlow, Apache Spark, and Google’s uncertainty-baselines. Based in San Francisco, he focuses on reliable A(G)I at scale and has a track record of improving model calibration, ELBO metrics, and multi-host, multi-device training pipelines.
12 years of coding experience
11 years of employment as a software developer
Doctor of Medicine (M.D.) Candidate (2/4 years), Doctor of Medicine (M.D.) Candidate (2/4 years) at The Brody School of Medicine at East Carolina UniversityBachelor of Science (B.S.) Computer Science, Bachelor of Science (B.S.) Computer Science at Appalachian State University
English
Github Skills (28)
spark10
python10
big-data10
machine-learning10
java10
scala10
javas10
keras10
deep-learning10
tensorflow10
bayesian-methods10
tensorflow-probability9
regularization9
ml9
n9
Programming languages (8)
Github contributions (5)
google/uncertainty-baselines
Aug 2020 - Dec 2022
High-quality implementations of standard and SOTA methods on a variety of tasks.
Role in this project:
ML Engineer / Data Scientist
ML Engineer / Data ScientistContributions:26 reviews, 107 commits, 3 PRs in 2 years 4 months
Contributions summary:Mike's commits primarily involve adding and modifying code related to training and evaluation of machine learning models within the "uncertainty-baselines" project. These changes focus on implementing metrics for various neural network architectures, specifically related to Bayesian Neural Networks (BNNs) such as Rank-1 BNNs. The user's contributions include the integration of metrics related to Evidence Lower Bound (ELBO) calculations and the expansion of existing metric aggregations.
implementationsstatisticsdata-sciencedeep-learningneural-networks
apache/spark
May 2015 - Dec 2015
Apache Spark - A unified analytics engine for large-scale data processing
Role in this project:Back-end Developer
Back-end DeveloperContributions:6 commits, 20 PRs, 129 comments in 6 months
Contributions summary:Mike primarily contributed to the Apache Spark project by enhancing the MLlib module. Their work involved making the `ChiSqSelector` class serializable and adding imports. Furthermore, the user updated the ML documentation by removing references to `fittingParamMap` and replaced them with `extractParamMap`. The contributions also included adding a `drop` function to the DataFrame, and adding a variety of methods to PySpark's distributed linear algebra classes.
analyticspythondata-processingsqlapache
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.