A key member of a team developing proprietary FLINT NLP Desktop + SAAS frameworks.- Implementing State-of-the-art ML + Feature-centered NLP solutions for authorship analysis. Automating complex scientific methodologies and statistical approaches to forensic linguistics, working directly with the key scientists in the field.- Working on a variety of tasks, from NLP and Database development to APIs and QA.FLINT Systems was founded to automate the process of Authorship Determination. It has developed its software platform to enable businesses, institutions, and law enforcement to analyze whether a certain document of questioned origin shares likely common authorship features with other documents of known origin applying standard linguistic rules and patterns. FLINT's technology creates a unique linguistic fingerprint and applies forensic linguistic methodologies to detect Contract Cheating and AI generated text. Flint’s key areas of focus are: Authorship Attribution, Contract Cheating, and Threat Assessment. The system is a fully automated solution, which help dramatically reduce time and costs required for running custom analysis. The user can control key variables to ensure that the analysis fits the particular goals for individual use cases.

Working on freelance commissions.- Solved different NLP problems: Text classification, Text generation, Relation Extraction, Named Entity Recognition, Keyword extraction.- Created web scrapers and parsers.- Gave AI and NLP talks for non-technical audiences.- Created a multilanguage Telegram gatekeeper bot for a community of 300+ expatriate IT professionals.

Primary teams: Semantic Web + AI.- Developed a Structured document module: Table ingestion (pandas), triplet generation (RDFlib), full-text search (embeddings + numpy for cosine similarity; common metrics such as Hamming and Levenshtein distance, etc.)- Researched and developed State-of-the-art implementations for Named Entity Recognition, Relation and Term extraction, and Topic modeling. - Created RESTful APIs for the modules using Flask.- Benchmarked Python-based NLP frameworks (Huggingface, SpaCy, CoreNLP, Stanza, etc.) on tasks of interest.- Integrated SpaCy 3.0 and transformers into the NER module, improving quality by 20%.- Developed the first iteration of a load- and performance-testing system for NER.Secondary team: QA.- Modularized and refactored testing framework, speeding up daily workflow by 10%.- Completely reworked case split for parallelization, reducing the number of operations by 50%.- Advised a team of senior engineers on Python-related problems.

Laboratory of Algorithms and Technologies for Networks Analysis.- Conducted autonomous research in NLP and Computer Vision and authored three NLP papers in Scopus-indexed international conference proceedings;- Was a board member and a double-blind reviewer (NLP track) at six international conferences;- Participated in academic seminars and assisted in the organization of laboratory-sponsored events.

- An app / social network for finding the best prices of everyday goods, making retail more transparent.- Developed parsers for 5 company websites, increasing the amount of product data available to the app by 20%. These datasets allowed the app to be used in bookstores and drugstores.- Using SQL and NLP, consolidated three private datasets of LENTA, one of the largest Russian retail chains, creating a production-ready dataset of over 10.000.000 articles.

Higher School of Economics and Netcracker (Nizhny Novgorod) Joint Laboratory.- Researched and compared 5 different multilanguage corpora for QA and sentiment analysis tasks;- Researched SOTA NLP methods for dialogue systems, such as DeepPavlov.

Assisting for ‘Computer Tools for Linguistics Research’- Addressed student questions;- Reviewed tools and prepared presentations.