Prof Amit Kumar Shanti Swarup Bhatnagar Awardee, 2018

Professor Amit Kumar is currently Jaswinder and Tarwinder Chadha chair professor in the department of Computer Science and Engineering at IIT Delhi. He holds a Bachelors degree from IIT Kanpur and PhD from Cornell University. Prior to joining IIT Delhi in 2003, he worked as a member of technical staff at Bell Laboratories, New Jersey and has held several visiting professor positions at Microsoft Research India, IBM Research India and Max-Planck Institute, Germany. He has received INAE (Indian National Academy of Engineer) Young Engineer Award, INSA (Indian National Science Academy) Young Scientist Award, IBM Faculty Award and was Max-Planck-India partner group research fellow during 2005-09.

Professor Kumar has been selected for the Shanti Swarup Bhatnagar Award, 2018 for Mathematical Sciences, the announcement for which was made on September 26.

In an interview with Dr Vanita Srivastava he talks about his research and the challenges in the area

What does this award mean to you?

 

The Shanti Swarup Bhatnagar Award is one of the most prestigious awards in the category of Science and Technology in India. This will certainly motivate me to work with even more enthusiasm in my research area.

 

What is the research for which you have been awarded?

 

My research interest lies in the area of theoretical computer science, with emphasis on problems arising in scheduling, resource allocation, graph theory and clustering. Many of these problems are fundamental in nature and have been studied for many years (or decades). Often one needs to design algorithms which use novel heuristics, andformally prove that they give good solutions.

An example of a scheduling problem would be processing tasks in large data centers. Typically, a large data center can have hundreds of machines processing millions of tasks. In such scenarios, how should the tasks be processed in an optimal manner? Optimality could mean minimizing overall delay of tasks, or energy consumed by the processors, or some combination of the two. Further, the tasks may not be given in advance. They are revealed over time, but the algorithm has to make decisions which take all adversarial scenarios in account. These algorithms are called on-line algorithms, and my research has given novel results for many such settings.

My research has focused on similar resource allocation problems in graph theory. Here we want to connect a set of users or computers in a network in an optimal manner. Again one needs to adapt to settings when the entire input is revealed over time.

 

What were the major challenges that you faced in this research? What is the way forward?

 

One of the major challenges of working in India has been scarcity of good PhD students and collaborators. Fortunately, IIT Delhi has excellent faculty in the area of theoretical computer science, and I have tried to build collaborations with colleagues in other institutes.

Looking forward, we want to motivate more undergraduate students to take up research as a career. Any research area needs a critical mass of researchers. This is slowly happening in theoretical computer science in India.

 

Can you brief a bit about the Indo-US Virtual Networked Joint Center.

 

The Indo-US Virtual Networked Joint Center is a two year project intended to foster research collaboration in the area of algorithms between partner institutions, i.e., IIT Delhi, IIT Kanpur, Carnegie Mellon University and Duke University. We have been collaborating on several projects over past few years. This project will facilitate more exchange visits between faculty and students in these institutions.