Machine Learning Developer in Minsk, Minsk Region, Belarus

The project name is Darwin (called “Interaxo” for production). It is a web-based document management solution enabling collaboration across organizational and geographic boundaries. Users can create as many rooms as they wish and can structure the content itself, or use built-in templates. Users can preview many types of files using just the browser without any installed plugins. The project provides transparent MS Word integration allowing the user to create/edit any MS documents (preinstalled Office is required).

API Outsourcing is an industry leading provider of FAO outsourced services, including a best-of-breed accounts payable and workflow platform. The API transforms manual, paper-based accounts payable, document management, accounts receivable, and accounting service processes using its custom proprietary imaging and workflow platforms.

The CeilingWizard is a graphical tool developed as a signed applet allowing the user to design a curved ceiling system and transfer the design to an established USG coding system, such as the “Genetic Code” for Curvatura. The CeilingWizard supports scalable CAD files and creates and sends the reports generated as HTML documents.

The CQR maintenance application is a web-based Java application that is responsible for objects stored in the existing CQR database. It also provides high-level database maintenance routines, such as cloning/moving complex tree-like database data structures. The purpose of the application is administration of a corporate quality reporting (CQR) system. The application was written in Java J2EE and designed to use the IBM DB2 database in which the CQR data was stored. Communication with the database was done by means of JDBC. The application consisted of static HTML pages, dynamic JavaServer Pages, and control servlets. It was designed as a 3-tier application: a thin client (a browser), the application server (IBM WebSphere), and the database (IBM DB2).

I’ve developed a project for high-load processing of events (about 150 per sec) from RTB auction. Each event was taken from the AWS Kinesis stream, analyzed, and stored to the Amazon DynamoDb after that serialized as a JSON string to the Amazon S3 storage. About 6 graphics were developed for displaying in real time all statistics data.

Everything was developed using Amazon cloud services: Kinesis stream, DynamoDB, S3 storage, RDS (MySQL), EC2

Using the Stanford NLP framework, I have developed a couple REST services. The first one should significantly speed up extracting personal names and related job titles from web pages. The second service developed for extracting company finance KPIs from news articles (mergers, acquisition, revenue, etc.).