Prof Sunita Sarawagi: Helping machines make sense

Sunita Sarawagi Professor, computer science and engineering, IIT-Bombay

In a rapidly changing world where data meets intelligence, Sunita Sarawagi has been contributing with research, teaching and application of artificial intelligence (AI) in many technological ideas and processes that make the world easier for humans. As a professor of computer science and engineering at IIT-Bombay for the past 26 years, she has not only shaped minds but also institutions, founding the Centre for Machine Intelligence and Data Science at IIT-Bombay.

Her work has garnered high honours in the discipline, including the Infosys Prize for Engineering and Computer Science in 2019, and recognition as a fellow of the Association for Computing Machinery. Sarawagi’s influence extends beyond research, with numerous award-winning publications and leadership roles in major conferences and journals, such as ACM Special Interest Group on Management of Data, and Neural Information Processing Systems. 

Her journey began at IIT-Kharagpur, where she earned her bachelor’s degree, and then completed her PhD in databases from the University of California at Berkeley. Over the years, her work and research expertise have been used in some of the most prestigious institutions in technology and research, such as Google Research, Carnegie Mellon University and the IBM Almaden Research Center. Her current research interests are sequence models for text and time-series, domain adaptation, effective human intervention in learning, graphical models and structured learning.

Like many of her generation, Sarawagi says that she chose computer science because of her good JEE rank. “I gradually gravitated towards AI over 30 years,” she says. While working on data mining and cleaning, because of its relevance in data analysis, Sarawagi encountered conditional random fields (CRFs), which attracted her to “the beauty of statistical machine learning”. 

Sarawagi is a pioneer in building tools that help computers understand unstructured data, like text, images, videos and numbers that don’t follow a set format. In the past, computers mostly worked with neatly organised data, but as the world shifted, data became more and more unorganised. Sarawagi was one of the first to tackle this challenge with machine learning. Instead of using slow, manual methods that others had relied on, she introduced a smarter approach using a model called semi-Markov CRFs. This model helps computers break down and label sequences of words—for example, identifying names of cities or companies in a sentence.