Roman A Sukhanov | skhnv.com
Deep Learning Engineer | Data Scientist | Python, Java
∎ TECH STACK
math and table operations with: numpy, scipy, pandas, geopandas
experiments in: jupyter lab/hub, colab by google, datalore by jetbrains
visualization using: matplotlib, altair, plotly
classic machine learning via: xgboost, LightGBM, scikit-learn
neural nets with: TensorFlow, Keras
tuning hyper parameters with: optuna
track ML experiments in: MLFlow
pack result using: Streamlit and/or Docker
database experience: MySQL, PostgreSQL, SQLite, MongoDB
experience with cloud platforms: Oracle Cloud Infrastructure, AWS
Java experience: Core, JCF, Guava, Lombok
Java tests with: JUnit
Java GUI experience: Swing, AWT, JavaFX
Java Web framework experience: PlayFramework
∎ EXPERIENCE
Sr. Data Scientist at HQ Revenue, Berlin
02.2022 - 06.2022
Plan, organize and monitor AI research projects. Initialize new research projects, negotiate KPIs. Check new hypotheses, develop approaches (whole ML pipeline) for AI application in future projects
Program manager at Gazpromneft Science & Tech Center, St.Petersburg
09.2021 - 02.2022
Plan, organize and monitor AI research projects. Initialize new research projects, negotiate KPIs. Check new hypotheses, develop approaches (whole ML pipeline) for AI application in future projects
Sr. Data Scientist at Gazpromneft Science & Tech Center, St.Petersburg
07.2019 - 09.2021
Developed and supported AI projects for oil and gas:
1. developed inpainting and extreme inpainting approaches for seismic data (2D CNN, TensorFlow+Keras, regression task).
2. developed a well log prediction project based on seismic data (CNN, TensorFlow+Keras, regression task).
3. supported project for seismic fault detection (3D CNN, PyTorch, binary segmentation task), working on performance part (training and evaluation).
Also, I supported an outsource team that developed a super resolution approach for seismic data (CNN, PyTorch, regression task).
1. developed inpainting and extreme inpainting approaches for seismic data (2D CNN, TensorFlow+Keras, regression task).
2. developed a well log prediction project based on seismic data (CNN, TensorFlow+Keras, regression task).
3. supported project for seismic fault detection (3D CNN, PyTorch, binary segmentation task), working on performance part (training and evaluation).
Also, I supported an outsource team that developed a super resolution approach for seismic data (CNN, PyTorch, regression task).
Data Analyst at Gazpromneft Science & Tech Center, St.Petersburg
12.2015 - 07.2019
Developed a desktop application for estimation of depth map uncertainties (Java + JavaFX). Created mathematical and statistical models for building depth maps.
Created different mathematical and statistical models for geology, oil and gas and drilling.
∎ EDUCATION:
St.Petersburg Mining University M.Sc. in Geophysics
09.2007 - 06.2012
∎ CONFERENCES:
NIS GeoConference NIS Naftagas, Novi-Sad
2018
Machine Learning Summer School Skoltech, Moscow
2019
BigData 2020 Marche Polytechnic University, Ancona
2020
Data Science in oil and gas EAGE, 2020, online
2020
Data Science in oil and gas EAGE, 2021, online
2021
∎ PAPERS
Кросс-валидационная оценка параметров регресии при интерпретации сейсмических данных ProNeft
01.2017
Применение фильтра Собеля для тектонического районирования на примере юрских отложений одного из
месторождений ОАО «Славнефть-Мегионнефтегаз» ProNeft
03.2017
Restoration of Seismic Data Using Inpainting and EdgeConnect SPE
10.2021
∎ CERTIFICATES
English language certificate, B2 - Upper Intermediate EF SET
09.2021
Getting started with TensorFlow 2 Coursera, Imperial College
01.2022
Customizing your models with TensorFlow 2 Coursera, Imperial College
02.2022