Work
Skills
Programming languages: Python (academic and industrial projects). Matlab, R (academic projects). Elementary: Java (Coursera course), C++ (2 years in Bachelor).
Human languages: English (IELTS 8/10), Dutch (B1 completed), Russian (native).
Other skills: NLTK, HuggingFace, Stanford CoreNLP, Stanza. MySQL, MongoDB, neo4j, ElasticSearch, Whoosh. Keras, PyTorch. Jupyter Notebook. LaTeX, Git, Bootstrap, Flask, HTML, CSS, Javascript.
Experience
ML and NLP Engineer
November 2022 - now. Dedalus
clinalytix team. Information extraction and text classification for multilingual clinical texts.
Senior NLP Engineer
March 2021 - October 2022. Faktion
Semantic search, topic modelling, word sense disambiguation, query expansion and correction & text classification. (Dutch, government documents). Collected feedback from users, carried out research and development, implemented a demo frontend, maintained the backend, prepared microservices API which were incorporated into an analytics platform.
Text quality assessment: readability, fluency, coherence. (English, legal documents.) Research and development.
Knowledge sharing sessions on Bayesian deep learning and graph deep learning.
Deep learning for Question Answering (QA)
October 2017 - August 2018. DFKI MLT
I worked on non-factoid multilingual (including code-mixed) QA. I have also developed an interactive prototype that demonstrates inner representations of an LSTM model for a better understanding of the network decision making process.
-
Ekaterina Loginova, Günter Neumann. An Interactive Web-Interface for Visualizing the Inner Workings of the Question Answering LSTM. EMNLP 2018, System Demonstration. Brussels, Belgium, November 2018
-
Khyathi Raghavi Chandu, Ekaterina Loginova, Vishal Gupta, Josef van Genabith, Günter Neumann, Manoj Chinnakotla, Eric Nyberg, Alan Black. Code-Mixed Question Answering Challenge: Crowd-sourcing Data and Techniques. ACL 2018, Third Workshop on Computational Approaches to Linguistic Code-switching. Melbourne, Australia
-
Ekaterina Loginova, Stalin Varanasi, Günter Neumann. Towards Multilingual Neural Question Answering. 1st International Workshop on Artificial Intelligence for Question Answering, Communications in Computer and Information Science (Springer). Budapest, Hungary, 2018
Master Thesis Internship
February - June 2017. Vrije Universiteit Brussel
I have
- organized corpus annotation for sentiment analysis (comments to “The Guardian” articles on climate change)
- extracted linguistic features and tested their influence on performance
- trained machine learning models (Support Vector Machines, Maximum Entropy, Naive Bayes)
- compared existing sentiment analysis tools (Stanford CoreNLP, Vader, Pattern, Polyglot)
Teaching Assistant
April - July 2015. Higher School of Economics
Course Social Network Analysis (in English).
Topics: graph analysis; with the use of software for creating a graph of connections extracted from the social network account. Responsibilities: I compiled an up-to-date list of resources and prepared handouts for exercise sessions in R. I have also checked solutions and provided feedback to help students.
Summer Internship
2013. Higher School of Economics
Project VisualMath: I have implemented interactive Javascript visualizations of interesting examples for calculus theorems. The resulting illustrations were actively used on the lectures and have been included in a published version of lecture notes.
Tutoring
2013 - now.
Teaching math to middle- and high-school students, as well as linear algebra, calculus, statistics and differential equations to university students. Also tutored bachelor students on Python and machine learning.