RISHI ALLA

- Created text classification models to help automate transformation of raw data to relevant domain specific labeled datasets.

- Created pipelines that takes in uploaded documents (PDFs, Word documents, OCR'd scanned documents, etc.) and cleans it to be used in NLP tasks.

- Used NLP methods and packages (spaCy, BERT, Hugging Face transformers, etc.) to extract metadata (titles, keywords, and summaries) from uploaded documents to custom search engine.

- Use fuzzy matching to find similarities between entries in multiple datasets to create a master dataset with all relevant information within.

- Decoupled hard-coded data from investment tool and implemented functionality which allows users to select data through a GUI which queries into a SQL Database.

- Improved upon existing Plotly Dashboard by making it more user friendly allowing users not well versed in code to utilize the investment tool.

- Created visualizations to view differences in portfolio weights and constraints and improved existing data visualizations to make it easier to understand for all users.