Education

Bachelor of Computer Science (B.S.)

Specialization: Machine Learning

• Automating the report generation from employee engagement surveys, saving 20 man-hours per report and enhancing timely decision-making.

• Design and develop course assignments and coding materials, improving student engagement and understanding
  of the course material.

• Developing an ETL process by establishing naming conventions for efficient survey data extraction and implementing a robust pipeline for data profiling, statistical analysis, sentiment analysis, and topic modeling, and automating the storage of survey results, pioneering data flow and analytics at Deutser.

• Facilitates students’ comprehension of financial concepts in the Quantitative Finance course taught through
  hands-on instruction in Python.

• Design and develop course assignments and coding materials, improving student engagement and understanding
  of the course material.

• Led the creation of a start-up as the lead developer, taking charge of product design and development, including setting up the database and establishing a streaming data pipeline.

• Customized a proprietary ESG taxonomy and ontology used to label 50+ company textual reports and 100+ news articles used for building an ESG knowledge meta-model that identifies differences between a company’s ESG reports and media coverage.

• Presented the start-up to potential investors in more than 20 presentations, securing $50,000 grant from NSF.

• Built an information extraction graph-based machine learning model using PyTorch that successfully extracted textual and positional information from 50+ different types of insurance documents.

• Developed a python pdf insurance document generator for 10+ insurance invoice types to enrich the dataset for training the model.

• Developed a python pdf insurance document generator for 10+ insurance invoice types to enrich the dataset for training the model.

• Built a complete ETL pipeline that includes bi-weekly retraining with Airflow on an Amazon S3 bucket from files that are entered by the users to the applications, which increased accuracy of the model by 1%.

• Analysed and extracted data from 10,000+ cancer tissue data points and their characteristics, such as gene mutation and expressions, to create a personalized drug recommendation system using collaborative filtering.

• Deployed a client application using Flask and React, which displayed the results of the model with data visualization techniques and a dashboard with user login functionalities.

• Extracted, translated, preprocessed and performed sentiment analysis with both English and Macedonian sentiment models and topic modeling over time on 200+ poems of Blaze Konevski, a famous Macedonian writer.