Bohumir Zamecnik
AI Researcher at Rossum
Chiang Mai, Thailand
Summary
👤
Senior🎓
Top SchoolBohumír Zámečník is an AI researcher and seasoned software engineer with 15 years of experience, currently driving OCR and ML performance at Rossum where he focuses on GPU profiling, model tuning and scaling production pipelines. He combines deep theoretical background in signal processing and computer graphics with practical big-data and backend skills—Scala/Java, Spark/Hadoop and on-prem GPU cluster ops—to turn prototypes into production-grade systems. A frequent contributor to Keras and the TensorFlow ecosystem, he’s improved multi-GPU training and addressed CuDNN integration issues, reflecting hands-on expertise with core deep learning tooling. His earlier work spans generative audio models, music signal processing and large-scale clickstream search, revealing an uncommon blend of audio research and PB-scale engineering. Comfortable leading teams or diving into code, he balances rapid prototyping with long-term maintainability and is skilled at squeezing 100x speed-ups out of pipelines. Based in Chiang Mai, he pursues woodworking and music projects outside work, demonstrating a practical, craft-oriented approach to complex problems.
15 years of coding experience
6 years of employment as a software developer
CourseraGymnazium ZamberkBc., Computer Science, Bc., Computer Science at Charles University in PragueMgr., Computer Graphics, Mgr., Computer Graphics at Univerzita Karlova v Praze
Czech, English, German, Spanish, Thai, Japanese
Stackoverflow
2,640reputation
645kreached
24answers
3questions
Github Skills (19)
python10
machine-learning10
git-repository10
rnn-model10
n10
keras10
deep-learning10
tensorflow10
github-repos10
cudnn10
testing8
passwords6
gradle6
scala6
java6
Programming languages (12)
Github contributions (5)
keras-team/keras
Dec 2016 - Aug 2021
Deep Learning for humans
Role in this project:
ML Engineer
ML EngineerContributions:11 commits, 10 PRs, 93 comments in 4 years 8 months
Contributions summary:Bohumir primarily contributed to the TensorFlow-based Keras deep learning library, focusing on enhancing its functionality and resolving critical issues. Their work involved implementing features for TensorFlow's K.Function() related to fetches and feed_dict, including unit tests, and adapting the library to support CuDNN-based GRU and LSTM layers. The user also addressed bugs, such as fixing resource leaks and resolving CuDNN and TimeDistributed conversion issues, demonstrating a deep understanding of the library's backend and its integration with different hardware configurations. Additionally, the user corrected documentation typos and upgraded dependencies.
pythondata-sciencedeep-learningneural-networksmachine-learning
bzamecnik/salutr
Sep 2012 - Nov 2014
Contributions:69 commits in 2 years 1 month
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.