Probed CLIP for racial bias in facial recognition during zero-shot classification and linear probe classification. Tested CLIP's understanding of primitive concepts by comparing accuracy after targeted image perturbations and masking.

Served as a teaching assistant for six different course staffs. Duties included managing a staff for a course with 250+ students, developing modules and presentations on the social impact of machine learning for three different courses, and facilitating discussions about data ethics with a non-technical audience.Socially Responsible Computing TA / Undergraduate TA - CSCI1460 (Computational Linguistics)Head TA - CSCI1470/2470 (Deep Learning)Undergraduate TA - CSCI1460… Show more Served as a teaching assistant for six different course staffs. Duties included managing a staff for a course with 250+ students, developing modules and presentations on the social impact of machine learning for three different courses, and facilitating discussions about data ethics with a non-technical audience.Socially Responsible Computing TA / Undergraduate TA - CSCI1460 (Computational Linguistics)Head TA - CSCI1470/2470 (Deep Learning)Undergraduate TA - CSCI1460 (Computational Linguistics)Undergraduate TA - CSCI2470 (Graduate-Level Deep Learning)Undergraduate TA - DATA0080 (Data, Ethics, and Society)Socially Responsible Computing TA - CSCI0220 (Discrete Structures and Probability) Show less

Lead the Brown branch of AIRES, relaunching the club to focus more extensively on engaging non-technical audiences with issues of AI fairness and regulation.

Designed, trained, and deployed a model to help optimize seller allocations given a set headcount, using gradient-boosting trees to measure potential change in client value in upcoming plan cycles. Conducted data discovery, feature engineering, and EDA. Developed tools to speed up SQL queries by over 70% and work across separate databases and schemas.

Implemented a client prioritization model for Red Hat, serving as a liaison between non-technical stakeholders and a data science team. Conducted EDA and produced data visualizations to communicate technical details to stakeholders and performed feature engineering for the model.

Leveraged fundamental analysis and ESG data to analyze energy and utility companies, creating recommendations that would be used to invest a portion of Brown's endowment.