Data Science Intern
Boston, Massachusetts, United States
- Developed an end-to-end data pipeline using Python, Pandas, and FAISS to normalize and deduplicate 600,000+ company names to create a directory of companies with unique IDs and short, clean names- Utilized advanced text preprocessing techniques, including stop word removal and hashing vectorization, to enhance data quality and improve the accuracy of company name matching and similarity scoring.- Built an app to dynamically search the company directory for user-inputted names, providing real-time updates and user options for ID assignments based on found matches.- Designed a Python script to interact with the GPT-4o API for generating GraphQL queries, simplifying vessel database querying by enabling natural language search capabilities for users rather than searching via filters- Enhanced the query generation process by integrating a GraphQL schema, example queries based on top searched fields, and hierarchical ship type data from a CSV file, allowing for more precise and context-aware database searches