• Achieved 25% improvement in entity extraction and search, leading to increasing conversion by more than 25% and reducing bot confusion by as much as 10%

• Designed, implemented, and maintained automated content generation systems, incorporating QA and style transfer using LLM pipelines. This streamlined onboarding for new clients by 87%.

• Developing a Hybrid (Intent + Intentless) chatbot by integrating LLMs like GPT3.5/4 with Rasa chatbots, resulting in a contextual and personalized chat experience for users. Demonstrated a reduction in confusion by over 50%.

• Worked on policy compliance in the Indian securities and exchanges sector, leading to 2 publications at WWW’22.

• Achieved over 97% accuracy in automated violation predictions by collaborating with legal professionals and fine-tuning a transformer model, posing the task as a multi label classification problem.

• Funded by JP Morgan Faculty Awards, 2021 & 2022

• Catalog Enrichment: Worked on developing an end-to-end pipeline to improve search queries by enhancing the catalog through customer reviews. Implemented multiple neural Aspect Based Sentiment Analysis papers.

• Language Enrichment: Worked on extending language support for 15 new languages. Extended the duckling parser for these languages to extract price information from user queries in real-time. Utilized Open Multilingual Wordnet to support automatic synonym enrichment for multiple languages. Developed a transformer-based system that supports multilingual word sense disambiguation.

• Analytics: Built WordGraph, a tool to visualize relations between search queries to help merchants make data-driven decisions, was featured in Forbes.

Persuing Research in NLP under Dr. Manish Shrivastava.

Areas of interest include:

• Events and Timebanks for Indian Languages
• Narrative Understanding