Two CS doctoral students recently received prestigious industry PhD Fellowships that recognize and support innovative research. Sangeetha Abdu Jyothi was one of 13 students worldwide who earned a 2017 Facebook Fellowship, and Silu Huang was one of 10 students who earned a 2017 Microsoft Research PhD Fellowship.

A member of CS Associate Professor Brighten Godfrey’s group, Jyothi is part of a research team that is designing a nearly speed-of-light Internet Service Provider (ISP). Current ISP networks use buried fiber optic cables to move data and run much slower than the speed of light in air.

According to Jyothi, the latency or delay is about 3-4 times higher in conventional fiber-optic networks than the speed of light latency because of physical limitations of fiber and geographical constraints of laying the fiber along the shortest routes.

To address this, Godfrey’s team is building a network or ISP primarily based on line-of-sight wireless links, meaning they’ll combine the speed of microwave connections—like those used to connect the stock exchanges—with the information carrying capacity of fiber. This will significantly improve performance of Internet applications.

As a 2017 Facebook Fellow, Jyothi will receive funding for conference travel and a $37,000 annual stipend for the next two years; Facebook will also pay her Illinois tuition and fees during that time. “The fellowship is a huge boost for my project,” said Jyothi. “I’m grateful to my advisor, Brighten Godfrey, too, because he has provided such great support for my work.”

A third-year PhD student, Huang is working with CS Assistant Professor Aditya Parameswaran’s group on the DataHub project, which is like a GitHub for structured data. DataHub will enable data scientists to better collaborate and analyze their shared data.

DataHub is increasingly necessary as data scientists create hundreds or even thousands of versions of the same data set while performing their analyses. Typically, a data scientist will copy an original data set, make changes or correct errors, and then place the new versions on the same shared file system, which results in massive redundancy and makes it difficult to share the newly constructed data sets.

To fulfill the DataHub vision, Huang has been studying how to adapt a traditional database for versioning. Specifically, she is developing data models and a partitioning algorithm to efficiently and effectively bolt-on versioning for relational database. Known as OrpheusDB, the hosted platform incorporates the strengths of a traditional database—consistency and querying with standard SQL commands—with new attributes like multi-version control and compact storage solutions.

In the summer of 2015, Huang worked at Microsoft Research on approximate query processing, which is a problem that arises when trying to extract aggregate information over massive data like sales revenue of individual products from a big database. She helped develop sampling and indexing techniques that enabled database users to quickly get a reliable answer to their query with guaranteed error bound. According to Huang, Microsoft is using this approximate query processing method internally. 

As a Microsoft Research PhD Fellow, Huang will receive funding for conference travel and a $28,000 annual stipend for the next two years; Microsoft will also pay her Illinois tuition and fees during that time.

“This fellowship means a lot to me because industry finds my work quite useful,” said Huang, who expressed appreciation to her U of I advisor Aditya Parameswaran, CS Professor Saurabh Sinha, and Microsoft advisor Bolin Ding (PhD ’12). “This gives me motivation to continue my work and allows me to be more focused on the DataHub project.”

Huang is the second CS student to receive this fellowship in the last 10 years. Chi Wang (PhD ’14), a former member of Professor Jiawei Han’s group, was a 2011 Microsoft Research Fellow.