Michael Karchevsky, Developer in Batumi, Adjara, Georgia
Michael is available for hire
Hire Michael

Michael Karchevsky

Verified Expert  in Engineering

Image Processing Developer

Batumi, Adjara, Georgia

Toptal member since September 12, 2016

Bio

Michael has a strong background in machine learning and AI, with diverse research experience in industry and academia. He has successfully led development cycles and has expertise in working with various data types. Michael has strong leadership, teamwork, quick learning skills, and a results-oriented approach. He excels under pressure and can meet challenging deadlines, and these qualifications make him a valuable candidate for positions in ML and AI.

Portfolio

DemandBase
PyTorch, Generative Pre-trained Transformers (GPT)...
Zarplata.ru
PyTorch, Python, Generative Pre-trained Transformers (GPT)...
Aurteen, Inc.
Image Processing, Python, PyTorch, TensorFlow, Deep Learning, Computer Vision...

Experience

  • Image Processing - 11 years
  • Computer Vision - 9 years
  • Generative Pre-trained Transformers (GPT) - 6 years
  • Deep Learning - 6 years
  • Python - 6 years
  • Natural Language Processing (NLP) - 6 years
  • Machine Learning - 5 years
  • Amazon Web Services (AWS) - 4 years

Availability

Part-time

Preferred Environment

PyTorch, Machine Learning, Data Science, Artificial Intelligence (AI), Python, Deep Learning, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Computer Vision

The most amazing...

...was the development of a text embedding system for a job board. This system yielded significant benefits in terms of recommender and search engine performance.

Work Experience

Senior Machine Learning Engineer

2021 - 2022
DemandBase
  • Managed a full-stack machine learning development cycle from research to deployment.
  • Performed data extraction from text sequences by extracting specific pieces of information or data from a larger body of text.
  • Developed deep learning models using machine learning techniques to train artificial neural networks to perform tasks.
  • Maintained the API in production, which involved ensuring that APIs function properly and efficiently in a live environment.
Technologies: PyTorch, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Deep Learning, Python, Machine Learning, Data Science, Named-entity Recognition (NER), Amazon Elastic MapReduce (EMR), Amazon S3 (AWS S3), Amazon EC2

Data Science Team Lead

2019 - 2021
Zarplata.ru
  • Directed a team of four data science and two data engineer professionals.
  • Led research and problem formalization (recommendation and search system, NLP, Churn, CV).
  • Oversaw the development of machine learning models.
  • Coordinated the deployment and maintenance of APIs in production.
Technologies: PyTorch, Python, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Recommendation Systems, Named-entity Recognition (NER), Computer Vision, Agile, Deep Learning, Management, Big Data, Machine Learning, LightGBM, Lighting, Artificial Intelligence (AI), Analytics, IT Deployments

Senior Data Scientist (Deep Learning)

2018 - 2018
Aurteen, Inc.
  • Supervised the research and the design of deep learning models.
  • Produced medical image segmentation and classification.
  • Developed complex production pipelines, from training to deployment.
  • Researched and implemented state-of-the-art data science techniques.
Technologies: Image Processing, Python, PyTorch, TensorFlow, Deep Learning, Computer Vision, IT Deployments, Medical Imaging

Data Scientist

2017 - 2018
SeatCrawler
  • Researched and developed new machine learning algorithms.
  • Implemented data crawlers and ETL.
  • Developed and improved data science pipelines.
  • Established a fault-tolerant multiprocessing workflow.
  • Maintained a Python web service (front-end + back-end), improved API integration, and developed new features.
  • Developed prediction of ticket prices and ticket sales.
Technologies: Bottle.py, Luigi, StatsModels, Scikit-learn, SQL, Pandas, Python

Big Data Analyst

2016 - 2017
P2 Energy Solutions
  • Automated a database migration pipeline from RDS to Redshift.
  • Built a new data warehouse to serve analytic operations.
  • Designed and implemented SQL scripts for data analysis.
  • Maintained and modified Spotfire data visualization pages.
Technologies: Amazon Web Services (AWS), Python, Spotfire, TIBCO, PostgreSQL, Redshift, Amazon EC2, Relational Database Services (RDS), Amazon S3 (AWS S3)

Software Developer

2013 - 2017
StreamData
  • Implemented a detection system and tracking of people on video, to help determine statistics in the supermarket.
  • Created software for automatic data processing and visualization of results.
  • Created an algorithm for a data science predictive model.
  • Developed image analysis algorithms for medical equipment.
  • Created neural network architecture and an API for styling images.
Technologies: PyTorch, Natural Language Toolkit (NLTK), Python, OpenCV, TensorFlow, Keras

Teacher of IT and Computer Science

2014 - 2016
Novosibirsk State University
  • Created an education methodology. Created lectures for fast understanding of key aspects. Structured programming knowledge.
  • Approved communication and presentation skills. Learned to transfer thoughts and ideas to students.
  • Got two students for scientific advisory as a result of acceptance of good training methodologies.
  • Trained students to be the best at programming, as measured by the results of the final exams.
Technologies: Image Processing, Python, OpenCV, Git, Object-oriented Programming (OOP), C++

Junior Researcher

2014 - 2015
Baker Hughes
  • Gained experience in fluid dynamics of wells and Monte Carlo methods.
  • Created optimization methods for solving the inverse problem (based on the Monte Carlo algorithms).
  • Implemented detailed statistical analysis of experimental data of drill bits.
  • Created an electromagnetic device model to predict sensor values.
  • Implemented 30+ data processing scripts (with full visualization).
Technologies: Mathematica, Wolfram, Python, OpenCV, C++

Laboratory Research Assistant

2009 - 2014
Novosibirsk State University, Institute of Thermophysics
  • Developed and modified algorithms for particle image velocimetry (PIV) and particle tracking velocimetry (PTV).
  • Implemented deep mathematical image processing algorithms.
  • Gained experience in analyzing algorithm documentation and using it as instructions for implementation.
  • Participated in many conferences, authoring or co-authoring thirteen scientific papers.
  • Created utilities and plugins for data processing.
Technologies: Tecplot, Linux, D3.js, OpenCV, Python, C++

Intern

2013 - 2013
Schlumberger
  • Modeled hydraulic fracturing.
  • Developed numerical data analysis algorithms.
  • Gained experience in computer simulations of physical processes, mathematical modeling, and numerical analysis.
  • Implemented visualization and preparation of technical presentations.
Technologies: D3.js, Petrel, Python, C++

4th PIV Challenge

http://www.pivchallenge.org/
I participated in the international scientific competition in data processing for the PIV and PTV algorithms. In the course of this work, I have created, implemented, or improved about 20 algorithms. Technologies used for development were C, C++, Python, OpenCV, OpenMP, machine learning, and MATLAB.

Home Credit Default Risk

https://www.kaggle.com/c/home-credit-default-risk
554th (top 8% out of 7198) - Home Credit Default Risk Kaggle challenge.

TGS Salt Identification Challenge

https://www.kaggle.com/c/tgs-salt-identification-challenge
I am proud to have achieved 27th place, ranking in the top 1% out of 3,234 participants, in a Kaggle competition focused on image and data processing. This competition was particularly challenging as it aimed to improve the accuracy of seismic imaging and 3D renderings by identifying salt deposits beneath the Earth's surface. The competition was hosted by TGS, a leading geoscience data company, who sought to leverage the power of machine learning to automate the identification of subsurface targets, ultimately creating more accurate and safer drilling practices for oil and gas companies.

PLAsTiCC Astronomical Classification

https://www.kaggle.com/c/PLAsTiCC-2018
52nd place (top 5% out of 1094) in a time series and data processing competition.

Russian Engineering Competition

I won the Russian Engineering Competition, 2015. Technologies used for development were C, C++, Python, OpenCV, and machine learning.

N+1 Fish, N+2 Fish Data Science Competition

https://www.drivendata.org/competitions/48/identify-fish-challenge/
11th place in Video Processing N+1 Fish, N+2 Fish competition (source data - videos).

Planet: Understanding the Amazon from Space Data Science Competition

https://www.kaggle.com/c/planet-understanding-the-amazon-from-space
222nd place (top 24%) - Planet: Understanding the Amazon from Space competition (source data - images)

Toxic Comment Classification Challenge

https://www.kaggle.com/c/jigsaw-toxic-comment-classification-challenge
406th place (top 9% out of 4551) in a text processing competition.

Avito Demand Prediction Challenge

https://www.kaggle.com/c/avito-demand-prediction
131st place (top 7% out of 1917) in a text, image, and data processing competition.

2018 Data Science Bowl

https://www.kaggle.com/c/data-science-bowl-2018
130th place (top 4% out of 3634) in automated nucleus detection.
2015 - 2017

Masters's Degree in Data Analysis

Yandex School of Data Analysis - Moscow

2014 - 2016

Postgraduate Degree in Physical and Technical Information Technology

Novosibirsk State University - Novosibirsk

2012 - 2014

Master's Degree in Automation of Physical and Technical Research

Novosibirsk State University - Novosibirsk

2008 - 2012

Bachelor of Science Degree in Automation of Physical and Technical Research

Novosibirsk State University - Novosibirsk

Libraries/APIs

Pandas, NumPy, SciPy, Keras, Matplotlib, Natural Language Toolkit (NLTK), OpenCV, Scikit-learn, SpaCy, PyTorch, NetworkX, SQLAlchemy, TensorFlow, Python Imaging Library, D3.js, Luigi, Bottle.py, Beautiful Soup, Java Digital Image Processing (DIP), Standard Template Library (STL), OpenMP

Tools

Seaborn, Gensim, Scikit-image, Git, Plotly, Mathematica, StatsModels, Jupyter, IPython, PyCharm, Tecplot, Amazon Simple Email Service (SES), Google Analytics, JetBrains, IPython Notebook, Spotfire, Apache, SQLiteManager, Named-entity Recognition (NER), Amazon Elastic MapReduce (EMR), ImageJ

Languages

Python, Multiscale Modeling Language (MML), SQL, Wolfram, Java, UML, C++

Paradigms

Functional Programming, REST, Distributed Computing, Object-oriented Programming (OOP), Parallel Computing, Concurrent Programming, Data-driven Programming, Model-driven Engineering (MDE), Agile, Management

Platforms

Jupyter Notebook, Amazon Web Services (AWS), Linux, Amazon, Amazon EC2, Petrel

Frameworks

Scrapy, Flask, LightGBM

Storage

PostgreSQL, Redshift, MySQL, JSON, SQLite, NoSQL, Amazon S3 (AWS S3)

Other

Data Science, Computer Vision Algorithms, Pipelines, Data Processing, Neural Networks, Deep Learning, Algorithms, Analytics, Image Processing, Data Analysis, Computer Vision, Mathematics, Machine Learning, Scientific Computing, Artificial Intelligence (AI), Predictive Modeling, Time Series, Text Classification, Text Categorization, Sequence Classification, Extractive Question Answering (EQA), Text Generation, Leadership, Bokeh, Tesseract, Physics Simulations, Statistics, Data Structures, Numerical Methods, Network Programming, Relational Database Services (RDS), TIBCO, Natural Language Processing (NLP), Recommendation Systems, Search, 3D Image Processing, Patterns, Time Series Analysis, Programming, Physics, Big Data, Lighting, IT Deployments, Medical Imaging, Images, 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