- Search engine for web (text indexing, storage, ranking)- SKU classification for marketplace aggregator

Performed research and prepared PoC for HR analytical tool. It allows to automatically estimate candidate’s job application.Suitability between resume and job position was derived based on semantic similarity metrics between them along with other’classic’ machine learning and information retrieval approaches. Achieved classification ROC AUC score about 0.82Technologies: Python, Scikit-learn, Pandas, XGBoost, word vectors, Flask REST API, Gensim, Docker.

Developed business-logic components for text documents collaboration platform.Technologies: Kotlin, PostgreSQL, Java, Spring Boot, REST API

Developed information extraction services for question answering engine in financial markets domain.Technologies: Java 8, Apache UIMA, REST API, Spring Boot, DKPro, LingPipe.

Developed and implemented logic for voice personal assistant bot.Developed classification module for for question-answering system.Developed information extraction system for financial markets search engine.Technologies: Java, Apache UIMA, Weka, Python.

Developed core part for virtual assistant chat-bot: chatting, question answering, online quizzes.Developed docker images for crawling and text analytics.NLP consultations for text analytics projects.Technologies: Java, Python, Apache Nutch, PostgreSQL, Docker.

Developed UI-testing automation framework, implemented modules of business logic and integration with third-party services for internal products of large transport company.Technologies: Linux, JavaEE, Spring, Apache Tomcat, Maven, JUnit, TestNG, Selenium, Vaadin, PostgreSQL.

Developed natural language processing modules for information retrieval systems.Implemented named entities recognition, synonyms extraction, clustering, classification, information extraction.Examples of projects:- Parsing internal streams of resumes and job postings and matching them against each other in order to find most suitable candidates for vacancies. Extracted professional skills and a lot of auxiliary information from job postings and CVs and measured their semantic relatedness using automatically acquired and manually supplemented skills ontology. Candidates were ranked with respect to level (years of experience) and “actuality” (how long time ago it was applied) ofeach their skill.- Generating HP products thesaurus extension for search engine of hp.com website. Trained HP products recognizer and created dictionary of synonyms and acronyms for HP products than can have more than single name (e.g. “HP Server Automation software” = “Opsware Server Automation System” or “HP ProBook s-series” = “HP Minis”).- Parsing of thousands of papers from internal HP Conference through last several years in order to explore dynamics in research activity of different RD departments.- Building fact extraction system for Russian language. Generating and manual creating rules for extraction people names, their affiliation to different organizations and job positions.- Hierarchical classification of user complains to HP hardware technical support.- Adaptation of Apache Mahout K-Means clusterer for internal HP artificial intelligence platform.Technologies: JavaSE, JavaEE, Python, Apache UIMA, Weka, LingPipe, Mallet, Open NLP, lots of NLP frameworks andtools.