Pavel Vychuzhanin Email & Phone Number

Belgrade, Serbia

Improving AutoML aspects for industrial failure predictions.- Refactoring and redesign of model for online air leakage detection. The model has been improved 1.5x by quality and 2-2.5x by execution time- Performing time-series analysis of industrial customer data. Providing data insights to the backend team- Implementing prototypes of ML pipelines for time-series classification industrial cases- Improving drawbacks of the existing ML pipelines- Conducting applied research on ML pipelines robustness and prediction qualityStack: time-series, machine learning, AutoML, python, scikit-learn, AWS, Ray

St Petersburg, St Petersburg City, Russia

Starting from a software architect and the main software developer I later became a team leader of AutoML framework development - Fedot.In less than 6 months our small team (8 people) successfully developed a PoC that showed competitive results in modern AutoML benchmarks.- Design software architecture, technologies stack, and API from scratch based on Python ecosystem- Planning, design and prioritize new features of the product- Coordinate researchers and developers in the team- Code review and refactoring the features implemented by team members- Teaching and mentoring junior developers the basics of object-oriented design, patterns, solid principles, and unit testing- Design and performing the experiments, preparing scientific papers and conference talksStack: AutoML, machine learning, python, scikit-learn, optimization

St Petersburg, St Petersburg City, Russia

I participated in several science-intensive projects where I did both research and softwaredevelopment. In the projects, I was the main developer and responsible for the most of technicaldecisions during the development of PoCs. The selected finished projects and my impact:Anomaly Detector for Arctic Weather Simulation Data (Dec 2017 - July 2018)Developed a Computer Vision algorithm for anomaly detection in multidimensional data based onConvolutional Neural Networks. An algorithm was wrapped into a Python software package thatwas later shipped and integrated with the complex Arctic Weather Simulation system.There I conducted most of the experiments and development.Stack: keras, tensorflow, numpy, python, multiprocessingEvolutionary optimization package of oil debit model parameters (NDA)I designed and implemented a python package for parameters optimization of equation-basedmodel for an oil&gas company. A hybrid (gradient and evolutionary) algorithm for optimizationwas developed and tested.Stack: python, scipy, numba, numpy, pandasSystem of industrial time series classification for large energy company (NDA)My goal here was to design the architecture of the system, and implement a user-friendly Python CLI API for datasets generation, model fitting, and time-series predictions. In particular:• Integration with a customer cloud database in order to wrap routine tasks related to datapreparation• Implementing and optimizing SQL requests for time-series data• Time-series analysis, including outliers detection, interpolation, and also visualization• Experimenting with classification and clustering methods The developed system was delivered and deployed in the customer infrastructure and the proposed technology of time series classification was patentedStack: python, scikit-learn, pandas, SQL, Snowflake, REST API, Flask

St Petersburg, St Petersburg City, Russia

I was a backend developer in the experimental project, which focused on the development of a PoC-solution in the field of virtualization of telecommunication devices.The solution was implemented mostly in Java according to the microservice architecture.My routine tasks and responsibilities included the following:- Design, developing and refactoring new features of several Java Spring services and REST API;- Writing unit, integration and contract tests using JUnit, Mockito, and Spring Cloud Contract;- Contributing to internal GitLab repositories according to Git Flow principles;- Synchronization with the team via Scrum-daily meetings and communication using SlackStack: Java, Spring, Microservices, REST API, Git, Scrum

Doctor Of Philosophy - Phd, Computer Science

Currently, I am a second-year PhD student in Computer Science. The research interests are automl, evolutionary optimization, data-driven.

Master'S Degree, Computer Science

I was a researcher in Natural Systems Simulation group: https://itmo-nss-team.github.io/ The research focus was on the development of new.

Bachelor'S Degree, Information Technology