Rudolf Eremyan

Using natural language processing algorithms with a combination of time series analysis approaches developed model for earlier trend detection in the Twitter stream.
Developed scripts for pulling and analyzing Twitter Stream using Twitter API.

Visualized results of the analysis with different plots for better interpreting.

Attribution modeling is the method used to measure the monetary impact a piece of communication has on real business goals, for example, sales, customer retention, revenue, and profit.

During working on this project I have extensively used SQL for data manipulation and analysis, as well as Python and R libraries. I have developed data migration and client notification scripts. Also, implemented data integrity tests for checking completeness and the correctness of existing data. Worked with an international team distributed around the world.

Using Google BigQuery analyzed news big dataset.

Trained machine learning models for text classification which used in text filtering mechanism. Integrated cloud ML services such as IBM BlueMix and Google Cloud NLP with an existing application.

Ti-Bot, the first ever Chat Bot to speak Georgian.

News article grouping tool uses word vectorizing technologies with a combination of clustering algorithms for automatically grouping similar articles parsed from news websites.

Social media sentiment analysis tool is a combination of natural language processing technologies and machine learning algorithms for predicting the sentiment for comments and posts, collected from social networks such as Facebook and Instagram.

Spell checker tool uses classical algorithms with a combination of powerful machine learning and natural language processing methods for detecting and correcting mistakes in the sentences. This product used by the largest companies in Georgia for detecting and correcting mistakes in documents.

Cryptocurrency prices monitoring tool uses time series analysis algorithms and Tweeter API combined with NLP tools such as Sentiment analysis, for monitoring and predicting price movements of Bitcoin and other cryptocurrencies.

A tool used for automatic identification of the Georgian dialects in documents from different sources such as forums, social networks, etc. It's based on machine learning classification methods and NLP approaches. During development, I worked with a group of linguists who prepared training and evaluated data for a classification model.

This project was awarded the "Best Scientific Research of the Tbilisi State University 76th Student Conference"

Developed a web application for storing, manipulating, and analyzing linguistic data.

Worked with clients team building new database for the pharmaceutical industry, by collecting, cleaning and managing data from different sources. Used AWS services for implementing ETL, storing logs, etc.

Parsed and analyzed a large volume of athletes' performance data. Applied the Monte Carlo statistical approach on athletes' performance data for simulating track and field competitions. Used AWS cloud services for running computations and storing generated results.