Archbase - innovative AI project in the field of architectural design of residential buildings based on multi-agent deep reinforcement learning methodAchievements:- Designed a universal, flexible software architecture that allows the business to globally change requirements, and all components of the system to adapt as quickly as possible. The implementation of new business requirements takes from 2 hours to 2 days of development- Optimized the code basing on performance tests and profiling (50x acceleration)- The code uses almost 100% of the machine's 60 vCPUs, and does not idle. This service is simultaneously used by neural networks on several super-computers with Tesla V100 video cards- Collected requirements and developed a service with real-time visualization of experiments for ML engineersTasks:- Creation, refinement and orchestration of microservices, building data models- Software design through diagrams using design patterns and UML notations, model C4- Finding out complex bugs both on the Java and Python side, and at the infrastructure level- Deploying microservices, configuring CI/CD, container registry, deploying the gitlab server, etc- Building a highly loaded simulator architecture for Reinforcement Learning- Development of a mathematical engine for working with geometryTechnology stack:- Java version 11/17 (JCF, concurrency, lambdas, functional interfaces, stream api, generics)- Kotlin (coroutines, inline functions, SAM interfaces, null safety)- Python (pytorch, asyncio, pytest)- Spring (Boot, Data JPA/JDBC, Security, Web, Actuators)- AMQP/RabbitMQ, REST- PostgreSQL, MongoDB- JMH, JUnit (>85% test coverage), mock objects, in-memory/embedded databases (h2/flapdoodle)- JMX, JFR- Maven, Git, Gitlab pipelines- YouTrack, Notion, Miro, Agile (SCRUM, KANBAN)- Drawio, PlantUML- Linux/Unix (bash, ssh/scp, vim, cron, curl, top/htop, etc.)- Docker, Yandex Cloud

Loudplay - cloud gaming service that provides a powerful virtual PCAchievements:- Developed a service for configuring hardware resources of a cloud PC via the XEN hypervisor- Participated in the development of the architecture of the cloud PC finite state machine (from power-on to shutdown state)- Built a clear logging system for microservices, on the basis of which many bugs were identified- Developed an Android package for connecting physical remote PC control devices (mouse, keyboard, gamepad, etc.)- Designed the MVVM architecture of the Android application, configured the DI and Jetpack components from Google. Link to the application: https://play.google.com/store/apps/details?id=io.loudplay.android.appTasks:- Creation, refinement and orchestration of microservices, building data models- Software design through diagrams using design patterns and UML notation- Finding out and fixing bugs based on data, collected from more than 1 million users, in DBMS, analytics and crash analytics services- Led the Android development team, defined the technical stack, conducted code reviews and interviewsTechnology stack:- Java version 8/17 (JCF, JNI, concurrency, stream api, generics), Vavr (functional programming)- Kotlin (coroutines, extensions, inline functions, SAM interfaces, null safety)- Spring (Boot, Data JPA/JDBC, Security, Web, WebFlux, Actuators)- REST, SOAP, XML-RPC- MySQL, Redis, Neo4j, MyBatis- JUnit, Mockito, Yandex Tank, Espresso- Swagger, Postman, Kibana- Gradle, Maven, Git, Github Actions- Jira, Confluence, Agile (SCRUM, KANBAN)- Linux/Unix (bash, ssh/scp, vim, cron, curl, top/htop, etc.)- Docker, Kubernetes, Helm, Yandex Cloud

Project work on the case for “Frisbi”Achievements:- Implemented and delivered to the customer an Android application using 3 neural networks on the GPU of the device (for segmentation of the main object, segmentation of digits inside this object and their classification)Tasks:- Implemented native C++ code via JNI- Translated neural networks from pytorch format to ONNX, and then to MNN for maximum acceleration of work (especially on weak devices)- Designed the MVVM architecture of the Android application, configured DI and Jetpack components from GoogleTechnology stack:- Java version 8 (JNI)- Kotlin (coroutines, extensions, inline functions, SAM interfaces, null safety)- C++- MVVM, Dagger2, Jetpack, Clean architecture- REST- JUnit, Mockito, Espresso- Gradle, Git- Asana, Agile (SCRUM)

Summer university internship in the data analysis department of X5 Retail groupAchievements:- My team and I closed the set goals and made a final presentation of our work, which we showed to the company's services managerTasks:- Writing scripts in the R programming language- Normalization of data, removal of outliers- Data clusterization- Setting the final score for each cluster

Android application for electronic document managementAchievements:- Implemented the functionality of signing a document using the SOAP protocol- Designed the data caching structure in the absence of the Internet- Implemented a built-in file viewer in the application with the ability to scroll and expand- Implemented an optimal search window (minimum network calls)Tasks:- Application development from scratch in a team of scrum master, tester, 2 android developers and team lead- Optimization, bug fixing, refactoringTechnology stack:- Java version 8- Clean Architecture, MVP- REST, SOAP, Retrofit- Room, SQLite- RxJava2, Dagger2