Kuznetsov Alexander Email and Phone Number

Responsible for developing machine learning practice in the company. Working on visual attribution for e-commerce catalog. Goods attribution extraction from catalog images using deep learning. Helping and consulting other teams in machine learning problems/project.

Working on search enhancement for one of the EPAM clients and consulting other teams in search, web scraping, machine learning, etc. Created a Solr indexer on spark. Done improvement in wildcard query: wild query expansion based on term frequency, most frequent comes to a search query, in Lucene in current version was based on alphabetical order.

Start-up in IT recruitment field. Responsible for backend architecture: search engine built on top of Solr, custom ranking for candidates accordingly to its skills, experiences, culture fit to the company. Autocomplete for education, company, skill and location fields. Matching and deduplication of candidates from different sources for example from Linkedin and Github, based on information in user profiles like job history, education, location and skill set. High-performance async web crawler for gathering information from the internet.

Proof of concept for using Solr for search in e-commerce:Working on complex spellchecking allowing correct mistake when slitted (food proccess or -> food processor) and merged (towellholder -> towel holder) words present in search queries.Stemmer that correctly working for case like this running -> run, men -> man.Microsoft technical computing:Working on implementing mathematical functions for Microsoft technical computing library. Helping team to found estimates on computation errors for functions.Prepare internal 2 day training on GigaSpace technologyCrawler for Ajax sites based on FireFox browser:Creating a crawler which can collect information about products from Ajax sites. For one site it was one backed which synchronize work of several crawler. A crawler was resented as FireFox browser with special plug-in which can emulate user behavior. For each site it was need to create a configuration which tell worker how crawling site. Both workers and back-end automatically deployed on Amazon cloud.Responsible for architecture, management of developers team. Also was core developer in the team.Products advertisement matching:Matching of products from different site via it titles. Using Lucene as search framework to find match, and Weka as library for classification is match correct or incorrect.Responsible for architecture and was core developer in this project.Terracotta module:Adding functionality for Terracotta module for Quartz, a schedule framework in java. After this adding failed job can be restarted in Terracotta cluster environment. It was a short project for one developer. Get expertize in java byte code manipulations.Proof of concept in memory data grid field:Study opportunity for using of in memory data grid in fraud detections. This project involve in competitions of two vendors in IMDG (Gigaspace and Oracle Coherence) and Oracle database. Was a one from developer team.

Web developer. PHP, Perl, Ruby on Rails, Javascript.