● Compared millions of rows of data across multiple different datasets using record linkage and fuzzy matching techniques to identify inconsistencies in text data to generate information leads. Identified missing data gaps that inform future data collection efforts.● Created a machine learning pipeline that processes billions of data points using scikit-learn and optuna and automatically tunes and deploys itself based on updates to its underlying data. ML model achieved high precision with… Show more ● Compared millions of rows of data across multiple different datasets using record linkage and fuzzy matching techniques to identify inconsistencies in text data to generate information leads. Identified missing data gaps that inform future data collection efforts.● Created a machine learning pipeline that processes billions of data points using scikit-learn and optuna and automatically tunes and deploys itself based on updates to its underlying data. ML model achieved high precision with medium recall and integrated into a Flask web app. ● Designed an AI search vector store for millions of vectors which compressed data storage requirements by 60x and increased search speeds by at least 100x.● Developed a machine learning model that identifies anomalous and synthetic data in datasets.● Refactored existing audio search capabilities and increased audio translation and transcription speeds by 300% by integrating a faster implementation of a language translation/transcription model Show less