About

Romal Thoppilan is a Founding Researcher at Character.AI, currently working as an Engineer at Google Brain. With over 5 years of relevant experience, Romal is an expert in AI, research science, language models, and infra. He has a Master's degree in Statistics from North Carolina State University and a BE Honours in Computer Science from BITS. At Google, Romal works on making the largest generative language model in Google safe, factual and human-like. He has led the efforts to rank homework problems with learning graph entities and solve AI-based tutoring with large language models in his previous role at Google Education. In his role at YouTube, he helped build reliable model training and serving infrastructure to actively classify videos with brand safety labels at scale. He also built systems to forecast performance metrics like user reach and impressions for any hypothetical ad campaign settings. Romal is a part of the team that developed LaMDA, a next-generation AI for dialogue. He has written a blog post about LaMDA and contributed to its press coverage. He has also researched and written about Meena, a conversational agent that can chat about anything. Romal is passionate about creating safe, grounded, and human-like dialog applications. In his LinkedIn summary, Romal quotes Niels Bohr, stating that an expert is a man who has made all the mistakes which can be made in a narrow field. This quote reflects his approach to learning and growing in his field of expertise. Romal is a dedicated researcher and engineer who is always looking for ways to innovate and improve.