I finished my studies at the joint master's program of MIPT and Schlumberger. During this program, I worked full-time at Schlumberger and used my projects' classical machine learning methods and neural networks for two years.
My daily stack:
- PyTorch & Keras & scikit-learn & XGBoost
- NumPy & Pandas & Matplotlib
- Git
Also I'm familiar with SQL, Docker, MLFlow, NLTK, Transformers, OpenCV and Catboost.
I'm never stop learning something new and I think we shouldn't be afraid of new tools and challenges.
- Help to mark quality of Internet resources about IT to improve the quality of search results. This helped make IT search results more relevant to the request.
- Responsible for reviewing the team's initial ratings, independent page ratings,
and the final verdict.
Achievements:
- Checked about 300 different pages about IT
- Made about 1000 conclusions about the quality of pages
-Processed a high-frequency distributed acoustic signal obtained
experimentally;
-Conducted research on standard supervised learning methods for this problem
(linear and logistic regression, decision trees, random forest and boosting);
-Developed an approach using neural networks to determine the sand
production and its localization. The idea was, as in speech recognition
techniques, to represent a high-frequency signal in the form of a mel-spectrogram.
Achievements:
-The models I developed have become part of the software for internal use and
are an alternative to expert assessment, as they work in close to real-time.