Projects: Industrial printing equipment producer Subject description: Implementing framework for maintenance log data from a huge network of industrial printing machines processing based on Kafka cluster in AWS with backup in s3. Starting from setting up development and production environments with GitLab CI pipeline, to implementing framework, HTTP API, UI with Sales Force and implement use cases in Flink which predict future problems, generating findings about coming service time and printing units productivity, based on variables in the log files (speed, total sheets printed, time since last maintenance, etc.). Project assignments in brief: Creating portable containerized developer environment with Docker and CI pipeline GitLab Setting up printing units data ETL. Clean log data of various structures and push it to Kafka using customized Avro Serialization schemas Implement use cases in Flink which perform real time analysis on the log data. Write results to MariaDB database using JDBC Unit testing of Streaming applications and Integration tests of streaming pipelines Deployment on multi-host, multi-container Docker cluster on AWS Technologies: Java 8, Confluent 4.0.0 (Kafka), Flink 1.4, Flink CEP, JUnit, GitLab, AWS (S3, EC2), Mokito, Scala 2.11.12, Apache Avro, Maven 3, sbt 1.1.1, Docker, Jenkins, SonarQube, Intellij Post Subject description: Implementation of Cloud Real-time Tracking Platform on Kafka; Deployment of Micro services on Rancher Kubernetes Docker Platform. A real-time event-driven platform for tracking mail and parcels was implemented. The Kafka-streams based solution has a Micro-Service Architecture developed in Java 8 and deployed fully on a Docker Kubernetes cluster. Project assignments in brief: Kafka-Streams Micro-service implementation using Avro Schemas and Confluent Schema-Registry Installation of Rancher, Kubernetes on dedicated servers Confluent Enterprise Docker cluster with Zookeeper, Kafka Brockers, (Avro) Schema-registry, Kafka-connector and Control-center (Docker compose) and Rancher cloud Spring Boot, Spring Data DAO implementation for Database connection Use of OrientDB or MongoDB and persistent graph storage Technologies: Java 8, Confluent 3.3.1 (Kafka), Kafka-Streams, OrientDB, Elasticsearch 5, Spring Boot 1.5.0, Git, GitLab, Rancher, Docker, Docker-compose e-Commerce Subject description: Implementation of Classification Models and Neural Network Models for a Recommendation Engine Prototype. In order to predict and recommend the products a user will buy again, transactional (order) data was used to develop Machine Learning models for a Recommendation Solution. Anonymized data of customer orders over time was used to predict which previously purchased products will be in user’s next order. The approach has been to combine manually extracted features with learned Neural Network based features that are used for a classification model at the end. For the final classification a GBM (Gradient Boosting Model) was learned. The final basket is determined using resulting probabilities, and peeking the product subset with maximum expected F1-score. Project assignments in brief: Feature Engineering using Spark SQL by joining and aggregating order history data for user, order and product combinations Using Spark MLlib for training Gradient Boosting Classifier Optimizing the classification model by using cross-validation and f1-score Deploying the spark-implementation on Hadoop YARN Cluster and Spark Standalone Cluster Technologies: CDH (Cloudera Distributed Hadoop) 5.11.2, Apache Hadoop 2.6.0, HDFS, YARN, Impala 2.8.0, Hue 3.9.0, Apache Spark 2.2.0, Python 3.5, Supervised learning, GBM (Gradient Boosting Model), Scala 2.11, Linux, GitLab, Docker

Responsibilities: CPA-network system maintenance: UX/UI , REST API development, QA, functional extension, benchmarking, security fault observation, integration with external services. Facts: Registered users amount twiced during last six months. Achievements: Approx. 30 integrations were made during last six months. Advertisers help requests amount decreased for 20%.

Responsibilities: Participation in construction equipment (excavators, loaders, rollers) auctions in EU for further import to Russia. Transportation, customs clearance, delivery to buyer or giving for rent. Mantainance of construction machinery park. Facts: Up to 5 pieces of machinery in a park simultaneously. Deliveries from EU and China. Monthly turn over up to 50 thousand euro. Achievements: Supplied salt packing production with sorting equipment from China manufacture. Contract signed and construction processed on UralChem holding's factory "AZOT" (Nitrogen and ammonia producer).

Business-informatics, Business-informatics

Moscow Institute of Physics and Technology

National Science and Research University of Perm