Artyom Grebenyuk

Project development from scratch. Managing team of 7 backend-developers, 5 frontend-developers, devops, 3 QA Proud of making online-gaming service with high-load - 5k rps, 200k notifications per second. Distributed concurrent game states, 2k simultaneous online on one session. Responsible directly or indirectly for all tech processes involved in development and support. Responsible for task decomposition, code review, architecture, key features development, incident handling, monitoring, scalability, etc.

Stack: - Python, Numpy, Pandas, Numba, Clickhouse, Postgres, Nameko, Asyncio, RabbitMQ, SQLAlchemy, Kubernetes, Docker, Redis Trading Platform Development Platform support for analyzing trading strategies on historical data, trading in the table, stock trading Optimization and refactoring of the platform for a larger number of customers and the required speed of platform response Breaking down the monolith into services, reducing response time from 40-60m for hourly strategies to 1m

I was developing Yandex's Pulse - intranet tool which aggregates Yandex apps logs, such as battery consumption, CPU time, etc. I was doing frontend - SPA with React, Redux, Highcharts. It was pages with charts of someway aggregated metrics. On the other hand, I was doing backend. I've used python/flask for API, MongoDB/mongoengine, and clickhouse with raw SQL for data storage. Especially I work on performance improvement of queries to clickhouse and it's schema itself.

I was looking for users in extremely big search logs (>1TB per day) to promote them to use Yandex services. I was making hypotheses about which user behavior may lead to users will use service after promotion. I have reached 1-5% CTR on cold pushes like - "Hey you've watched Red, would you like to watch Red 2" or "It seems like you like action films about cars (I've used Yandex Object Search databases to find regularities in users' favors), I sure you will like Fast'n'Furious".

've developed an application that was used to dataset mark up. I've investigated, train, test, and integrate neural network to classify symbols on license plates. I've Improved the performance of license plate detection and symbols segmentation video-analysis algorithms. I was developing and supporting application that was used to connect and configure remote radar-systems. I was using C++/Qt for developing desktop applications, and python for some handful scripts and NN training.