Eric Battenberg
Software Engineer at Google DeepMind
San Francisco Bay Area United States
Summary
👤
Senior🎓
Top SchoolEric Battenberg is a software engineer with 14 years of experience specializing in sound understanding and machine perception, currently working on multimodal AI within the Frontier AI organization at Google DeepMind. He combines deep academic training (PhD, UC Berkeley) with industry research experience across Google Research, Baidu, and music-focused roles, delivering production-ready machine listening and speech systems. His open-source contributions include rigorous test automation for librosa’s CQT implementation and enhancements to GPU-accelerated libraries like cudamat and neural-net tooling in Lasagne, demonstrating attention to numerical correctness and reproducibility. Comfortable spanning research and engineering, he has a track record of shipping scalable audio models and improving core libraries that underpin audio analysis at scale. An underappreciated strength is his focus on edge cases near the Nyquist frequency and robust initialization checks—small fixes that materially improve reliability in audio pipelines.
14 years of coding experience
19 years of employment as a software developer
PhD Electrical Engineering and Computer Sciences, PhD Electrical Engineering and Computer Sciences at University of California, BerkeleyBS Electrical Engineering, BS Electrical Engineering at UC Santa Barbara
Github Skills (16)
neural-network10
cuda10
audio10
machine-learning10
error-handling10
librosa10
dspic10
python10
theano10
dspace10
numpy10
test-automation10
testing10
documentation9
linear-algebra9
Programming languages (7)
Github contributions (5)
Lasagne/Lasagne
Feb 2015 - Feb 2015
Lightweight library to build and train neural networks in Theano
Role in this project:
ML Engineer
ML EngineerContributions:16 commits, 4 PRs, 93 comments in 7 days
Contributions summary:Eric primarily focused on enhancing the `lasagne` library, a tool for building neural networks in Theano. Their contributions included refining the `norm_constraint` function, adding new functionalities, and optimizing existing code. The changes involved improving the flexibility and usability of the library's weight normalization and gradient clipping mechanisms. The user also addressed code formatting and documentation issues.
pythondeep-learningtheanoneural-networksdeep-learning-library
cudamat/cudamat
Jun 2014 - Aug 2014
Python module for performing basic dense linear algebra computations on the GPU using CUDA.
Role in this project:Back-end Developer
Back-end DeveloperContributions:10 commits, 1 PR in 1 month
Contributions summary:Eric primarily focused on improving the `cudamat` library by adding error checks and documentation. They added a `cublas_init` error check to the `test_cudamat.py` file and the `cudamat.py` file to ensure proper initialization. Furthermore, the user documented the `max_ones` argument to `cublas_init` and performed some code cleanup by removing tabs, blank lines, and unused imports. Finally, the user also adhered to PEP8 formatting conventions.
cudapythonlinear-algebragpu-accelerationmatrix-multiplication
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.