Owned prototyping features, analyzed synthetic user data to develop scalable MVPs to meet user needs.

Developed MLOps pipeline utilizing Kubeflow and Docker to continuously fine tune and redeploy open source machine learning models to increase context-specific generation.

Scraped real-time New York City Transit (NYCT) Twitter data to analyze service reliability at different train stations, classifying different delay types and assessing impact on rider travel time using ANN, linear regression, and SVM.

Performed sentiment analysis on customer replies to NYCT Tweets to evaluate emotional impact of planned and unplanned delay types.

Maintained a database and recommendation system based on natural language user input through PostgreSQL and Tensorflow.

Extracted keywords from user input using named entity recognition to structure database queries best suited for user requirements.

Improved and normalized database schema to ensure data integrity and optimize storage and retrieval performance.