Ben Lee
Staff Software Engineer at Google DeepMind
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolBen Lee is a Staff Software Engineer and former mathematician with 11 years of experience building statistical and distributed systems for ML and finance, currently working at Google DeepMind on Bayesian methods, generative models, and TensorFlow Probability. He has a track record of joining early-stage teams and scaling them into robust engineering organizations—most notably as first hire and later VP of Engineering at Determined AI and Head of Software at The Voleon Group—while also shipping core infrastructure like hyperparameter search, GPU cluster managers, and production trading systems. Comfortable straddling research and engineering, he contributes performance-focused fixes to high-profile open-source projects such as TensorFlow Lingvo (TPU embedding and MLPerf support). Based in San Francisco, he prefers individual-contributor research roles at mid-to-large companies but remains open to compelling opportunities, and brings both deep quantitative training (PhD-level math background) and hands-on ops experience managing compute clusters and production deployments.
11 years of coding experience
13 years of employment as a software developer
Bachelor's degree Mathematics, Bachelor's degree Mathematics at Harvard UniversityPhD Mathematics, PhD Mathematics at University of Chicago
Github Skills (9)
tensorflow10
tpu10
python9
distributed-computing9
deep-learning8
machine-learning8
natural-language-processing7
devops7
nlp7
Programming languages (4)
Github contributions (5)
tensorflow/lingvo
Nov 2018 - Nov 2022
Lingvo
Role in this project:
Back-end & DevOps Engineer
Back-end & DevOps EngineerContributions:2 reviews, 122 commits, 1 comment in 4 years 1 month
Contributions summary:Ben primarily contributed to the Lingvo project by implementing enhancements and fixes related to TPU (Tensor Processing Unit) embedding. Their work involved adding options to pin variables to CPU, supporting multi-task TPU embeddings, and resolving issues related to infeed threading and the use of tf.identity. Moreover, the user addressed a write race issue in beam search operations and integrated the support for the MLPerf benchmark, showing a focus on improving the system's performance and efficiency.
lingvospeech-recognitiontranslationspeech-to-textmachine-translation
benlee/dotfiles
Jul 2015 - Jul 2015
Contributions:1 push, 1 branch in 1 day
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.