Dmitriy Serdyuk
Senior Research Scientist at Meta
Montreal, Quebec, Canada
Summary
🤩
Rockstar🎓
Top SchoolDmitriy Serdyuk is a Senior Research Scientist with 11 years of experience bridging deep learning research and production-grade ML engineering, currently based in Montreal and working at Meta after recent roles at Google and DeepMind. He earned a PhD in Machine Learning from Université de Montréal and brings a rare blend of theoretical foundations (MSc in theoretical CS, BSc in solid state physics) and hands-on systems work. His open-source contributions to influential projects like Theano/PyTensor, Blocks, and pytorch-kaldi demonstrate expertise in core ML frameworks, model search/decoding algorithms, and robust data pipelines. Dmitriy has a track record of improving maintainability and performance—refactoring hyperparameter tuning tools, adding HDF5 dataset support, and implementing beam search primitives—showing attention to both algorithmic detail and engineering quality. He moves comfortably between research and engineering: publishing rigorous methods while shipping code that scales in real-world toolchains. Colleagues describe him as a pragmatic problem-solver who surfaces subtle reliability issues in data and model workflows before they become production incidents.
11 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy - PhD, Machine Learning, Doctor of Philosophy - PhD, Machine Learning at Université de MontréalMaster of Science (MSc), Theoretical Computer Science, Master of Science (MSc), Theoretical Computer Science at Saint-Petersburg Academic UniversityBachelor's degree, solid state physics, Bachelor's degree, solid state physics at Saint-Petersburg State Polytechnical University
Github Skills (29)
algorithm10
hdf10
pytorch10
algorithms10
data-pipelines10
python10
operation10
data-engineering10
data-set10
tensorrt10
machine-learning10
data-structure10
numpy10
datasets10
deep-learning10
Programming languages (17)
Github contributions (5)
mila-iqia/blocks
Dec 2014 - Sep 2018
A Theano framework for building and training neural networks
Role in this project:
Backend Developer
Backend DeveloperContributions:313 commits, 88 PRs, 66 pushes in 3 years 9 months
Contributions summary:Dmitriy contributed to the development of the beam search functionality within the Blocks framework for building neural networks. The commits focused on implementing the core structure and components of a beam search algorithm, including initial state computation, probability calculations, and sequence rearrangement. The changes involved modifying existing files like `blocks/beam_search.py` and `blocks/bricks/sequence_generators.py` to support the new search features.
pytorchdeep-learningtheanoneural-networksmachine-learning
mila-iqia/fuel
Feb 2015 - Sep 2018
A data pipeline framework for machine learning
Role in this project:ML Engineer & Data Engineer
ML Engineer & Data EngineerContributions:73 commits, 23 PRs, 12 pushes in 3 years 7 months
Contributions summary:Dmitriy primarily contributed to the development of a data pipeline framework for machine learning. Their work included implementing a new HDF5 dataset class for efficient data loading and integrating it with the existing framework. Furthermore, the user made the codebase compatible with Python 3 and addressed potential issues like handling missing HDF5 files, demonstrating a focus on data processing and model input.
pipelinedata-sciencepipeline-frameworkmachine-learningmlops
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.