Latest Content
I am a Computational Linguist/Data Scientist, and programmer who specializes in natural language data, with a focus on conversational AI (textual and spoken).
My current research leverages LLMs for sentiment, toxicity detection, and dialogue state tracking. In my previous role, I ran experiments toward improving our conversational AI in the areas of NLU with a focus on small talk, dialogue state tracking, and speech analytics. I was also active on the expansion of chatbots into the voice channel.
My dissertation research involved using machine learning models to analyze and interpret the importance of features used to predict sentiment and subjectivity in spontaneous, dialogical conversation. Prior research used classifiers to model pronunciation variation in connected speech.
Outside of research, much of my focus was in building a large corpus of translation pairs (over 1000 languages) which was sourced from print and online sources, cleaned, and validated with consideration of the native orthographies. This corpus is known as the UW Master Lexicon.