George Nigmatulin

George Nigmatulin

Machine Learning Engineer

Curriculum Vitae

Work expirience

Samsung R&D Institute Russia (Jul 2021 – Present, Moscow, Russia)
Machine Learning R&D Engineer / Sensor Solutions Team, Life Care Solutions lab.
  • Developed regression models for estimation BFM (Body fat mass), SMM (Skeletal muscle mass), ICW & ECW(Intracellular Extracellular water) by a multi-frequency signal (released in Samsung’s Galaxy Watch 4)
  • Designed system of Data Quality estimation, detection erroneous measurements and classifying user usage errors basedonOutliers’ Detection algorithms. Quality metric increased by 21% compared to heuristics
  • Developed Sweat loss estimation algorithm. Accelerometer-based Neural Network estimates running distance (MAPE=7.7%, R2=0.95and Polynomial Kernel Ridge Regression estimates the loss in ml (RMSEBWP=0.3%, R2=0.79).
Skolkovo Institute of Science and Technology (Aug 2020 – Jul 2022, Moscow, Russia)
Researcher / Laboratory of applied research Skoltech-Sberbank.
  • Developed an active learning algorithm modified by anomaly detection for planning experiments. The number of required labeling has been reduced by 58%, the quality of the regression model has been improved by 19%
  • Researched the uncertainty scores (total, data, knowledge) using a Bayesian ensemble of decision trees, and anomaly detection methods; researched correlation dependencies of uncertainty/abnormality scores.
Bank Orenburg (Jun 2019 – Jul 2020, Orenburg, Russia)
Analyst, Data Scientist.
  • Forecasted the ATM daily cash demands based on Exponential Smoothing, ARIMA, Neural Networks, SSA.
  • Developed a Discrete model of optimal cash management in ATM branches on forecasted cash withdrawals using Dynamic programming,which increased the profitability compared to the classical model by 30%.

Education

Moscow Institute of Physics and Technology (Sep 2020 – Jun 2022, Moscow, Russia)
Master’s degree in applied mathematics and physics / Machine learning and data analysis.
  • The Interfaculty Department of Information Transmission Problems and Data Analysis
  • Graduation project: “Anomaly detection aided Active learning on smart watches data”
  • Average grade 4.8 out of 5.0 for 2 years.
Orenburg State University (Sep 2016 – Jun 2020, Orenburg, Russia)
Bachelor’s degree in applied mathematics and computer science.
  • Department of Math and information technology
  • Graduation project: "Research of machine learning models for predicting and optimizing ATM service"
  • Average grade 5.0 out of 5.0 for 4 years.

Professional skills

Languages: English - Advanced C1, Russian – Native
Programming: Languages: C++ / C#, Python, R, Matlab, Matcad, Octave, SQL
DevOps: Linux, Bash, SSH, Git, DVC, Hadoop, Docker, AWS, Cookiecutter-data-science, WandB, MLFlow

Quantitative Research toolkit

ML packages: Scikit-learn, SciPy, LightGBM, Catboost, SHAP, Prophet, Numpy, Pandas, Matplotlib, Seaborn
Deep Learning: TensorFlow, PyTorch
Finance: Yfinance, Statsmodels, Tsfresh, FinancePy, Quantpy, Pyfolio, Backtrader, Quantstats, Interactive Brocker
Area of specialization: Machine learning, Deep Learning, Causal and Bayesian methods, Uncertainty quantification, Optimal control, System analysis, Quantitative Finance, Strategies, Predictive Modelling, Portfolio optimization.

Pdf version of Curriculum Vitae

https://drive.google.com/file/d/16Fcq8b-fn2SDC3vSC8bn-utplYZdkIih/view?usp=share_link