Dmitriy Bespalov
Senior Applied Scientist at Amazon Web Services (AWS)
New York, New York, United States
Summary
👤
Senior🎓
Top SchoolDmitriy Bespalov is a Senior Applied Scientist at AWS with 13 years of engineering experience and a PhD-backed foundation from Drexel University, specializing in machine learning, pattern recognition, NLP, recommender systems, and unsupervised anomaly detection. He has a strong production focus—shipping ML solutions using Python, PyTorch, Transformers, and scalable MLOps tools such as SageMaker, Glue, and Kubernetes. Dmitriy pairs research-grade rigor with pragmatic engineering: he contributes to open-source performance tooling (notably adding robust serialization and pickling support to the widely used hnswlib nearest-neighbor library) and implements C++/Python integrations via pybind11. Equally comfortable in backend systems and model development, he brings deep familiarity with approximate nearest neighbor search, retrieval-at-scale, and the full deployment lifecycle. Colleagues rely on him to close the loop from prototype to reliable, auditable production services.
13 years of coding experience
Bachelor of Science and Master of Science, Bachelor of Science and Master of Science at Drexel University
Github Skills (12)
nearest-neighbor-search10
deserialization10
serializable10
c-language10
approximate10
cprogramming-language10
serializer10
neighbors10
python10
pybind1110
nearest-neighbors10
serialization10
Programming languages (5)
Github contributions (5)
nmslib/hnswlib
Oct 2020 - Nov 2020
Header-only C++/python library for fast approximate nearest neighbors
Role in this project:
Back-end Developer
Back-end DeveloperContributions:10 reviews, 23 commits, 2 PRs in 1 month
Contributions summary:Dmitriy implemented new methods to serialize and deserialize the `HierarchicalNSW` class, adding functionality to save and load the index to/from streams. They also added several properties to the `Index` class in the Python bindings, enabling pickling and providing read-write access to the number of threads and read-only access to various index parameters. Furthermore, they refactored the serialization process, and verified knn_query results after pickling index objects and ensuring the results matched those of a brute force search. Finally, they added documentation to the python bindings.
python-libraryneighborspythonheadernearest
dbespalov/reeb_graph
Feb 2013 - Apr 2013
Contributions:41 commits, 1 comment in 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.