Miroslav Batchkarov
Research Engineer at Google DeepMind
Münster, North Rhine-Westphalia, Germany
Summary
👤
Senior🎓
Top SchoolMiroslav Batchkarov is a research engineer and seasoned software developer with 14 years of experience building machine learning and NLP systems, currently working at Google DeepMind after senior research and engineering roles at AWS. He combines deep academic training (PhD in Informatics) with hands-on production wins—at AWS he led speaker diarization and distributed ML efforts that achieved multi-fold speed, memory and error-rate improvements and substantial cost savings. As a founder and interim CTO he shipped commercial legal and financial information-extraction products and scaled small teams to deliver customer-facing systems on tight timelines. His open-source contributions include improving the usaddress ML parser (wrapping models as scikit-learn estimators and adding cross-validation) and hardening statsmodels' MultiComparison tests, reflecting a focus on reproducible, well-tested tooling. Based in Münster, Germany, he blends research rigor with practical engineering, often surfacing non-obvious infrastructure fixes (e.g., authentication and shared-library issues) that improve security and reliability across teams.
14 years of coding experience
12 years of employment as a software developer
Doctor of Philosophy (Ph.D.) Informatics, Doctor of Philosophy (Ph.D.) Informatics at University of Sussex
English, German, Russian, Bulgarian
Stackoverflow
15,777reputation
1.7mreached
282answers
33questions
neural-network
top-5%
nltk
top-5%
scikit-learn
top-1%
nlp
top-1%
python
top-1%
machine-learning
top-1%
Github Skills (31)
python10
data-science10
address-parser10
testing10
scikit10
statistics10
machine-learning10
parse10
natural-language-processing10
scikit-learn10
nlp10
pandas9
java9
hypothesis-testing9
nltk9
Programming languages (9)
Github contributions (5)
statsmodels/statsmodels
Mar 2014 - Aug 2014
Statsmodels: statistical modeling and econometrics in Python
Role in this project:
Data Scientist
Data ScientistContributions:9 commits in 5 months
Contributions summary:Miroslav primarily contributed to the testing and enhancement of the `MultiComparison` and related functions within the `statsmodels` library. Their work involved implementing unit tests to validate the behavior of the `group_order` parameter, ensuring correct ordering of results. Furthermore, the user expanded the docstrings for clarity and implemented improved error handling within the `MultiComparison` class. This indicates a focus on improving the robustness and usability of statistical analysis tools.
forecastingpythonregression-modelsstatsmodelsstatistics
datamade/usaddress
Oct 2014 - Oct 2014
Role in this project:ML Engineer & Data Scientist
ML Engineer & Data ScientistContributions:13 commits, 1 comment in 3 days
Contributions summary:Miroslav contributed significantly to the `usaddress` library, focusing on improving the machine learning model used for address parsing. Their work involved adding and refining model training parameters, including integrating cross-validation techniques for parameter tuning. They also wrapped the model in a scikit-learn estimator to facilitate easier parameter optimization through grid search. Further contributions addressed safety checks and the overall efficiency of the model and related tooling.
python-librarynlppythonstringsaddress
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.