Andrei V. Keino
Address: Zolotodolinskaia – 29 – 406, Novosibirsk, Russia
Email: andreikeino@gmail.com
Telephone: +7 953 763 9042
Skype: andrei_keino
Gender: Male
Nationality: Russian
Target position: Data Scientist, applied mathematician.
Examples of task solving:
Code example:
https://github.com/AndreiKeino/CSharp-like-properties-in-python
https://github.com/AndreiKeino/camex
Website example:
Algorithm examples:
https://vixra.org/pdf/2110.0094v1.pdf
https://figshare.com/articles/Valve_Spring_Fault_Detection_Final_1_pdf/7745330
Experience:
December 2021 – March 2022
MKK Dengimigom, Naberezye Chelny
Developer of data science algorithms
Developed web – service for scoring with linear dependency of distribution of recommended loan sum with known max/min, etc…
depending on the default probability with Flask, docker, Catboost
July 2021 – August 2021
Temporary distant job for some private entrepreneur, Kazan, algorithm developer.
Derived the equations and developed the code for estimation the probability and some statistics
for some multivariate distribution.
Software & languages & libraries used: Python, NumPy, MatPlotLib, sftp, ssh, bash.
December 2020 - February 2021
Global Monitoring, Orenburg, AI specialist, distant temporary job.
Developed the analytical algorithm that can detect fuel anomalies (refuel, drain, outliers)/ as I believe, much better than
one from here: https://gurtam.com/en/wialon
Software & languages & libraries used: Python, R, pandas, numpy, scipy, matplotlib, plotly, sftp, ssh, bash.
June 2019 - October 2019
Russian Thermotechnical Institute, Data Scientist, temporary job.
Implementation of a physical models, time series analysis and forecasting.
Development of the library for forecasting the malfunctioning of the power plant equipment.
Algorithms: Linear models, ARIMA, Dynamic harmonic regression, etc….
Software & languages & libraries used: Python, R, Postgresql, Influxdb, Pandas, NumPy, SciPy, Scikit-Learn, MatPlotLib, Plotly.
September 2016 – October 2018
NPP TIC, Perm, Programmer (research engineer indeed)
Functions: Planning and carrying out the experiments. Development of the DSP algorithms for vibration-diagnostics, utilizing some data science algorithms. The code for couple of the segmentation algorithms, wavelets (DWT, MODWT, CWT), noise reduction with MODWT, functionality for rotating machinery failure diagnostics with the Continuous Wavelet Transform (CWT), digital filtering, signal decimation, digital measurement of the phase difference with the Zero CRossing with Filtration method, etc…
Software & languages & libraries used – Matlab, C#, C/C++, Python, R, Pandas, NumPy, SciPy, Scikit-Learn, MatPlotLib, Plotly, Shiny, Accord.NET, …
April - June 2016 — I worked in ASF GS in Siberian Branch of Academy Of Science
Functions: Soldering of some hybrid(-circuit) boards, mounting and testing some of the experiment automation modules.
November 1996 – April 2016
About 11 last years I worked as a programmer, mostly as a freelancer (undocumented). Programming languages/libraries - C++; C#
13 July 1989 – 21 November 1996
I worked in the Institute of Thermophysics, Novosibirsk 6 years, made mostly hot - wire measurements, some vacuum evaporation also. Main duties – selection and development of the experimental environment, planning and carrying out the experiments, processing the experiment results. I’ve got 7 scientific publications on hot – wire measurements.
Skills:
- Programming experience: 12+ years
- Programming languages: - Python, R, Matlab/Octave, C#, C/C++
- Frameworks, tools, etc…: WinForms, PyQt/PySide2, Flask, Plotly.js, cvx, cvxpy, docker, microservices, scoring.
- Digital Signal Processing – Wavelets, filtering, etc…
- Data Science: - Statistics (hypothesis testing, ANOVA, regression analysis, nonparametric statistics), Principal Component Analysis,
regressions - Linear, Ridge, Lasso, Huber, Quantile, Logistic.
- Statistical learning – KNN, SVM, Random forest, Cluster Analysis.
- Deep learning: - FC, CNN, LSTM, GAN’s, Computer Vision, Pytorch.
- Time series analysis & forecasting – ARIMA, SARIMA, Time series decomposition (X11, SEATS, STL), Generalized linear models, Exponential Smoothing State Space Model, Regression with ARIMA errors, Dynamic harmonic regression, Hierarchical time series, Vector autoregression, Neural network autoregression, Bagging.
- Physics: - Thermophysics, Aerodynamics.
– Etceteras: Latex, Writing scientific articles.
Portfolio:
Portfolio optimization website - full – stack web development with Flask, Brython (frontend), Flask – Socketio, SqlAlchemy, nginx, uwsgi; website -
asset-master.net
- asset-master.net website highlights:
asset-master.net website can optimize portfolio using arbitrary user data in csv format
asset-master.net website routines uses daily stock data (not yearly or monthly)
asset-master.net website performs processor high - load math routines one by one
asset-master.net website shows task processing log in real time
Convex optimization, PyQT. Portfolio optimization (Markowitz model). Libraries: Cvxpy, PyQt.
https://pypi.org/project/camex/
Regression analysis and statistical testing:
https://rpubs.com/andrei_keino/402966
Hypothesis testing:
https://rpubs.com/andrei_keino/401944
FC neural networks with PyTorch (MPL):
Convolutional neural networks with PyTorch (ResNet implementation):
Recurrent neural networks, LSTM, attention with PyTorch:
Generative adversarial networks:
Forecasts: SARIMA:
Digital signal processing:
https://figshare.com/articles/Valve_Spring_Fault_Detection_Final_1_pdf/7745330
Website:
At https://www.w3schools.com/spaces/
https://andrei-keno-cv.w3spaces.com/index.html
Linkedin:
https://www.linkedin.com/in/andrei-keino-37334619/
EDUCATION:
1982 – 1989
Novosibirsk State University, Physics Department
Specialization: Aerodynamics and hydrodynamics
GPA 4.18
OTHER EDUCATION:
1992 – 1995
Novosibirsk Electrotechnical Institute, Post – graduate
Specialization: Thermophysics and thermodynamics
GPA 4.66
2018 - completed the Stanford StatLearning - SELF PACED Statistical Learning course.
2018 - Studied the Data Science specialization at Coursera – first 9 courses of 10 without getting the certificate, My Courseworks & Quizzes can be found here https://rpubs.com/andrei_keino , https://andrei-keino.shinyapps.io/developing_data_products_course_project/ .
2018 - Studied the Deep Learning specialization at Coursera without getting the certificate.
2019 - Studied the Bayesian Statistics: From Concept to Data Analysis course at Coursera without getting the certificate
2019 - Studied the Bayesian Statistics: Techniques and Models course on Coursera without getting the certificate. Capstone project is here: https://github.com/AndreiKeino/Coursera_Bayesian_Statistics_Techniques_and_Models/tree/master/Bayesian_Statistics_Techniques_and_Models/week_5
2019 – Studied the Practical Time Series Analysis at Coursera without getting the certificate. Quizzes are here: https://github.com/AndreiKeino/Coursera---Practical-Time-Series-Analysis/tree/master/Practical%20Time%20Series%20Analysis
2020 – Studied the course EECS 498-007 / 598-005 Deep Learning for Computer Vision: https://web.eecs.umich.edu/~justincj/teaching/eecs498/
Assignments are here:
https://github.com/AndreiKeino/EECS-498-007-598-005-Deep-Learning-for-Computer-Vision
2020 – Completed the courses EE364a, EE364b: Convex Optimization
https://see.stanford.edu/Course/EE364A
https://see.stanford.edu/Course/EE364B
Homework is here:
https://github.com/AndreiKeino/Convex-Optimization-EE364
2018 – Kaggle competitions. My Kaggle rating – top 5%. My Kaggle profile - https://www.kaggle.com/andreikeino .
LANGUAGES: Russian (native),
English (advanced; Speaking grade (1 – 5) 3; Reading grade 4)
List of Scientific Publications:
Simple and effective algorithm for constant state detection in time series.
https://vixra.org/pdf/2110.0094v1.pdf
Backtesting Investigation of Effect of the Optimized S&P 500 Portfolio Diversification with L2 Regularization.
https://vixra.org/abs/2105.0143 - Actually this is my first coursework on the “Convex optimization” course.
Detecting a Valve Spring Failure of a Piston Compressor with the Help of the Vibration Monitoring.
preprint:
https://figshare.com/articles/Valve_Spring_Fault_Detection_Final_1_pdf/7745330
Experimental and numerical modeling of the vortex furnace aerodynamics.
Russian Journal of Engeneering Thermophysics – 1996, N. 1.
Modelling the vortex furnace Aerodynamics.
4th European Conference on Industrial Furnaces And Boilers. Espino-Porto-Portugal, 1-4 April, 97. Preprints, Vol. 2
Hot – Wire measurements in three – dimensional turbulent flows with the probe rotated in one plane.
A. V. Keino, V. V. Salomatov.
Thermophysics and Aeromechanics. Vol. 4, N 1, 1997, Russian Academy of Science, Siberian Branch.
Исследование структуры потока в отрывном течении на треугольном крыле.
Известия Сибирского Отделения Академии наук СССР. Сер. Технических наук, 1989, вып. 3. с. 54.
Бардаханов С.П., Кейно А. В., Козлов В. В.
Исследование развития возмущений в области отрыва над треугольным крылом.
Сибирский Физико – Технический Журнал, 1993, вып. 6 с. 22.
Бардаханов С.П., Кейно А. В., Козлов В. В.
Исследование аэродинамики турбулентного потока в вихревой топке ЦКТИ.
Тезисы Регионального семинара ”Новые технологии и научные разработки в энергетике” Новосибирск, апрель 1995 г
Кейно А.В Саломатов В. В.
О термоанемометрических измерениях в трехмерных турбулентных потоках при повороте датчика термоанемометра в одной плоскости.
Теплофизика и Аэромеханика, 1996. № 2
Кейно А.В Саломатов В. В.
Participation in Physics Olympiads:
1980, 1981: Participating in regional Physics Olympiad.