Implemented and deployed GPS-based time-tracking, securing two new construction clients for the company

Maintained software on a day-to-day basis through quality assurance and unit testing for newer features

Preprocessed review data using SK Learn, NLTK, and pandas, converting user text into graph representations

Developed a Graph Neural Network architecture with PyTorch to classify argumentative structures accurately

Explored review helpfulness and developed a BERT-based architecture to surpass current SOTA benchmarks

Addressed the extraction of high-quality, argumentative tweets by framing it as a three-way text classification task

Converted models in existing literature to code and performed Deep Learning feature analysis using Captum.AI

Developed CNN & BERT models (PyTorch) with .81 F1 score and published a first-author paper at ArgMining’21