Tomas Shestakov

Senior Software Engineer

Coin Metrics•  October 2016 - Present

Developing a blockchain analytic platform in Kotlin that stores data of various digital assets in a uniform way in Postgres Database and Kafka. Working with full nodes API to obtain blockchain data. Calling Ethereum smart contracts. Made distributed lock implementation based on PostgreSQL Advisory Lock to add a sharding possibility. Reducing applications' memory consumption (16 times), and improving reliability (the number of Atlas-related incidents has decreased significantly). Non-obvious optimization of PostgreSQL queries. Parallel query execution to increase throughput. Create and analyze performance and resource consumption metrics. Deploying applications to the k8s environment.

System Architect

Okko•  March 2021 - October 2021

Architecture design of a monolith backend application transformation into microservices and migration into a Cloud ecosystem using a serverless application model. Development of the reporting system architecture for copyrighters based on Airflow, Python, and Pandas. Сoordinated the work of DevOps and programmers. Decreased response time of web services by using asynchronous Kotlin Coroutines and Flows. Fixing incorrect data schemas and wrong use cases when working with Cassandra reduced the load on the Cassandra database server.

Senior Java Developer

Pixonic•  December 2016 - March 2021

Working on a Big Data analytics system for War Robots mobile multiplayer game. It is a unique in the market schema-less, near real-time system capable of handling more than 1 billion events per day from mobile devices and servers. It has a custom SQL-like query language and a powerful user-friendly UI. There are integration libraries for Java, C#, iOS, Android, and Unity. I made significant contributions to distributing architecture, writing storage engine, and event generation services. Thanks to this, we were able to reduce the cost of ownership of the system and improve its performance.Using Apache Cassandra as events and intermediate state storage, Apache Kafka as a persistent message queue, Ansible for microservices deployment, Presto for complicated queries, Python (Pandas) for additional data analysis, and React for UI rendering. Partially made an R&D project to transfer the core functionality of the analytic system to Hadoop and Spark technologies.