• Owned end-to-end development of product features using Java, SpringBoot, Postgres, JavaScript, and Node

• 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

• Collaborated with Dr. Jon Park to assess the argumentativeness of product reviews using SOTA NLP techniques

• 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

• Led a team of 3 research assistants under Dr. Jon Park in NLP, focusing on Argument Mining from Tweets

• 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