(Last Updated: Oct 30, 2023)

I am an incoming research associate at the National Centre for Text Mining at the University of Manchester, where I work under Dr. Sophia Ananiadou exploring the scientific document understanding capabilities of large language models (LLMs).

I also currently collaborate with the Oxford TVG Lab (Prof. Philip Torr)and Meta researchers (Ser Nam Lim) on identifying and mitigating the potential for influence operations on social media using LLMs.

My past research experience includes working with the joint NLP Lab of IIT Patna and IIT Bombay under the guidance of Dr. Sriparna Saha and Dr. Pushpak Bhattacharyya. There, I work on a variety of NLP problems and collaborate with researchers from Microsoft, the University of Tokyo, and Amazon Alexa. More broadly, I worked on online safety, AI integrity, and user analytics. This encompasses projects such as developing advanced AI-powered content moderation techniques, testing the boundaries of LLMs, and implementing AI across sectors including education, law, and healthcare. My work has led to numerous publications in top-tier venues including EMNLP 2023, ACL 2023, CIKM 2023, ECML 2023, ACM MM 2023, ECAI 2023, ECIR 2022, and IJCNN 2022/2023.

My research focuses on developing techniques to ensure generative AI systems are safe, ethical, and socially responsible as their capabilities continue to rapidly advance. I am particularly interested in studying and mitigating the potential societal impacts of increasingly powerful generative models, with specific emphasis on the following aspects:

  • Mitigating misuse of generative models to spread misinformation or toxicity online..
  • Analyzing and reducing social biases inherited from training data.
  • Advancing interpretability of large language models.
  • I am also a core member of SimPPL, a research collective in the global south, developing open-access trust and safety tools for international orgs. including nonprofits and newsrooms. We aim to build better civic integrity tools such as Parrot Report. We’ve worked with Anti Defamation League, Tech Global Institute, The Sunday Times, Ippen Media, Deutsche Welle, and others on:

  • Study the spread of unreliable news and identify influence operations online (Parrot Report).
  • Deploying a WhatsApp chatbot for menstrual health education, fueled by local NGO insights and societal norms. (Sakhi Bot)