Dmitrii Murygin, Developer in Bishkek, Chuy Province, Kyrgyzstan
Dmitrii is available for hire
Hire Dmitrii

Dmitrii Murygin

Verified Expert  in Engineering

Software Developer

Location
Bishkek, Chuy Province, Kyrgyzstan
Toptal Member Since
December 7, 2021

Dmitrii is an experienced C++ developer and deep learning engineer. As a middle C++ developer he has five years of experience in research computation related projects. As a deep learning engineer he has two years of solid experience in NLP direction connected to transformer and BERT based models.

Portfolio

Innopraktika
C++, CMake, Clang, Google Test, Git, Ubuntu, Visual Studio Code (VS Code)...
Huawei Technologies Co.
Python 3, TensorFlow, PyTorch, BERT, Custom BERT, Neural Networks

Experience

Availability

Part-time

Preferred Environment

Ubuntu, Visual Studio Code (VS Code), CMake, Clang, Python 3, Vim Text Editor, C++

The most amazing...

...model I’ve developed is a BERT model in C++ that has the same performance on a mobile device as made using ONNX Runtime.

Work Experience

Research Software Engineer

2021 - PRESENT
Innopraktika
  • Developed a new method of 3D image segmentation based on supervoxel clusterization.
  • Developed a new method of pore-network extraction based on discrete Morse theory.
  • Implemented three parallel algorithms for 3D image segmentation and two parallel algorithms for 3D image filtering.
  • Implemented a cutting-edge programming module to simulate fluid filtration in pore-network models.
Technologies: C++, CMake, Clang, Google Test, Git, Ubuntu, Visual Studio Code (VS Code), OpenMP

Developer/Researcher

2018 - 2021
Huawei Technologies Co.
  • Developed a BERT model for mobile devices using C++. Compared it with other BERT versions converted from the most famous frameworks.
  • Made some contributions to various projects such as TensorFlow, Swift, LLVM, Stardust, and TVM.
  • Implemented the following neural network compression techniques: quantization, distillation, and prunning.
Technologies: Python 3, TensorFlow, PyTorch, BERT, Custom BERT, Neural Networks

Mobile BERT Benchmark

https://github.com/Vooblin/onnxbert
A benchmark for BERT I created that is converted by ONNX for mobile devices. It is possible to get performance for tiny, base, and large versions of the BERT model. To create a mobile BERT version, it uses the ONNX Runtime tool and creates a fused BERT model.
2014 - 2018

Bachelor's Degree in Mathematics and Computer Science

Moscow State University - Moscow, Russia

Libraries/APIs

TensorFlow, OpenMP, PyTorch

Tools

CMake, Git, Vim Text Editor, Open Neural Network Exchange (ONNX)

Platforms

Ubuntu, Visual Studio Code (VS Code)

Languages

C, C++, Python 3

Frameworks

Google Test

Other

Clang, Deep Learning, LLVM, Machine Learning, Natural Language Processing (NLP), BERT, Custom BERT, Neural Networks, Generative Pre-trained Transformers (GPT)

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring