Julian Shun concentrates in parallel computing as a principal investigator at CSAIL and assistant professor in the Electrical Engineering and Computer Science department at MIT. Shun said his goal is to design new parallel algorithms and programming frameworks to solve problems more efficiently.
Shun obtained his Ph.D. from Carnegie Mellon University and worked as a Research Fellow at UC Berkeley’s Miller Institute. Currently at MIT, he focuses on parallel computing in an effort to speed up important computations in the real world.
Shun said that due to limits in physical hardware, computations must take advantage of parallelism and utilize multiple resources to speed up tasks being solved. Speeding up computations is becoming more important because the data sets they are dealing with are particularly large, according to Shun.
Shun said that in order to extract useful information from large data sets and receive insights in a timely fashion, you need to have fast solutions to analyze the data sets. He compares parallel computing to the construction of a home to illustrate its efficiency.
“If you’re trying to build a house, you can hire one person to build the house or you can hire a group of people to try to build the house, and, usually, the group of people building the house would be faster,” Shun said.
However, there are various technical challenges that arise when designing parallel programs.
“Once you have multiple people building the house, there are challenges on how to divide the work so that everyone has something to work on and how to coordinate among the different people so that they don’t step on each other’s work,” he said.
Certain tasks have dependencies among one another, which have to be coordinated among different people. Parallel computing is about taking advantage of multiple resources to speed up something you are trying to do, according to Shun.
Currently, Shun’s main project aims to design fast parallel solutions for graph processing. This work has a myriad of applications in industry since graphs are tools to represent relatonships between objects of interest, particularly financial, transporation and pharmaceutical sectors.
Financial companies can use these tools to monitor transactions between people and detect fraudulent behavior in real time. Transportation companies can use graphs to represent distances between locations of interest on a map. Pharmaceutical companies can utilize these to represent interactions in protein or gene networks that arise in biology.
Shun said he is interested in learning more about applications of his work in companies, as well as trying out his solutions on large data sets that companies have, and exploring collaboration and funding opportunities.