Tobias Domhan
Senior Research Scientist at Google
Berlin Metropolitan Area Germany
Summary
🤩
Rockstar🎓
Top SchoolTobias Domhan is a Senior Research Scientist in multilingual NLP with 15 years of experience building and shipping ML systems across research and production at Amazon and Google. He combines deep research instincts with pragmatic engineering, having implemented core deep learning primitives (sigmoid layer in Caffe) and optimized RNN/GRU components in MXNet for high-performance training. Tobias has strong DevOps and backend chops from contributions to Sockeye and production ML tooling, enabling reproducible neural machine translation and stable GPU allocation. Based in Berlin, he bridges academic rigor (MSc Freiburg, exchange at UMass Amherst) with large-scale industry impact, and his career shows a pattern of moving inventions from prototype to production at cloud scale. An often-overlooked thread is his hands-on work across CPU/GPU implementations, highlighting fluency across low-level performance engineering and high-level NLP research.
15 years of coding experience
13 years of employment as a software developer
Master of Science Computer Science, Master of Science Computer Science at The University of FreiburgBachelor of Engineering Information Technology, Bachelor of Engineering Information Technology at DHBW StuttgartGraduate Exchange Student Computer Science, Graduate Exchange Student Computer Science at University of Massachusetts Amherst
English, German, Spanish
Github Skills (34)
hdf10
machine-translation10
pytorch10
c-language10
caffe10
neural-machine-translation10
python10
gpu-programming10
mxnet10
machine-learning10
recurrent-neural-networks10
encoder-decoder10
deep-learning10
deep-neural-networks10
neural-network10
Programming languages (12)
Github contributions (5)
awslabs/sockeye
Jun 2017 - Jan 2023
Sequence-to-sequence framework with a focus on Neural Machine Translation based on PyTorch
Role in this project:
Back-end & DevOps Engineer
Back-end & DevOps EngineerContributions:5 releases, 115 reviews, 146 commits in 5 years 7 months
Contributions summary:Tobias primarily contributed to improving the project's usability and maintainability through various commits. These included enhancing PyPI compatibility by updating instructions and modifying the setup file, mocking dependencies for building documentation, and splitting requirements files. They also improved the stability of the system through fixing the code and enhancing GPU allocation. Further contributions include parameter improvements for training and added git hash and version logging.
sequencetranslationsockeyeneural-machine-translationseq2seq
apache/mxnet
Mar 2017 - May 2018
Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more
Role in this project:ML Engineer
ML EngineerContributions:9 commits, 11 PRs, 76 comments in 1 year 2 months
Contributions summary:Tobias primarily contributed to the implementation and testing of Recurrent Neural Network (RNN) components, specifically a Gated Recurrent Unit (GRU) cell, within the MXNet deep learning framework. This involved modifications to existing RNN APIs, including adapting the GRU cell to the cuDNN version for performance optimization. Further contributions extended to testing and verifying the functionality of various RNN cell implementations (LSTM, GRU, and stacked cells) and ensuring the proper handling of argument types within the framework.
pythonschedulerdataflowmutationdata-science
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.