San Francisco Bay Area
Innovative leader in applied science and AI, specializing in developing scalable machine learning solutions. I have led cross-functional teams to drive advancements in Generative AI, Multilingual, and MultiModal Search capabilities. Passionate about leveraging cutting-edge technology to solve complex problems.
I teach Generative AI at UCSC NLP Program. The course explors the practical applications of deep generative models across diverse domains, including natural language processing, computer vision, reinforcement learning, and retrieval augmented generation for information retrieval.
Managing a cross-functional team of Applied Scientists, Engineers, and Managers, to deliver transformative innovation through Generative AI at Amazon's AGI. I lead the development of scalable machine learning systems that power multilingual Question-Answering systems, Knowledge Graph & Reasoning and MultiModal Search capabilities across the Amazon AGI
• Led the development and productization of novel NLP algorithms for IBM Watson's Natural Language Understanding (NLU) cloud service for advanced text analytics. • Designed proof-of-concept and prototypes for a novel explainable framework to improve customer retention. • Developed various compression techniques to improve inference time of machine learning models leading to cost savings for production systems. • Collaborated with research and product teams across multiple geographies to introduce new and improve existing NLP algorithms for IBM's NLU cloud.
- Developed techniques to recognize the specific arguments and counter arguments people tend to advance and group them across discussions as Facets in Social media conversations. - Developed a systematic approach to leverage pyramid based summarization framework to identify central arguments in social media dialogue. ( NAACL 2015) - Developed a novel method for ranking arguments in conversational data. ( Semdial 2017) - Introduced a new task of Argument Facet Similarity (AFS), and created a novel corpus of dialogic arguments aimed at identifying various facets across opinionated dialogue. (SIGDIAL 2016) - Developed topic-independent features for classifying (dis)agreement in online dialogue. ( SIGDIAL 2013)
Developed a novel antonym dictionary based method for sentiment analysis in customer service data.