I am a researcher at IBM IRL in Bengaluru, specializing in various areas of natural language processing. Prior to joining IBM, I gained valuable experience working for Verisk Analytics, KPMG, and Interactions LLC over a span of five years. I earned my Ph.D. from IIIT-Hyderabad in 2017 and subsequently pursued a postdoctoral fellowship at the University of Colorado, USA.
Throughout my career, my research has focused on several key areas, including conversational AI, document understanding, natural language understanding, and information extraction and verification. I have authored over 30 research papers on these subjects, which have been published in reputable journals and conferences.
Visual Document Understanding:
Developing end-to-end information extraction models for visually-rich documents (forms, invoices, insurance slips).
Incorporating textual and visual modalities for effective information extraction.
Conversational AI:
Developing natural language understanding models for virtual assistants.
Tasks include slot filling, entity grounding, and entity resolution.
Enabling accurate understanding and response to user queries and commands.
Natural Language Generation:
Developing neural generation models based on encoder-decoder and decoder-only paradigms.
Tasks include document summarization, paraphrasing, question generation, and information extraction from noisy documents.
Natural Language Inference:
Developing fact-verification and extraction models for semi-structured data.
Analyzing behavior when subjected to controlled probes.
Syntactic Parsing:
Creating treebank datasets.
Developing parsing systems for noisy and curated data in Indian languages.