June 19, 2020

Spotlight session with PhD students from the 2019 Facebook Fellowship cohort

By: Meta Research

Each year, we invite PhD students from around the world to apply for the Facebook Fellowship, a program designed to encourage and support promising doctoral students who are engaged in innovative and relevant research in areas related to computer science and engineering.

As a continuation of our regular Fellowship spotlight series, we’re highlighting Aishwarya Ganesan (University of Wisconsin–Madison), Kartik Hegde (University of Illinois at Urbana-Champaign), Maria Bauza Villalonga (Massachusetts Institute of Technology), and Neil Gaikwad (Massachusetts Institute of Technology). From teaching robots to identify objects to developing human-centered AI and public policy for sustainable development, these four Fellows showcase a wide array of research areas.

Improving the reliability of distributed systems without compromising performance

Aishwarya Ganesan is a PhD candidate in computer sciences at the University of Wisconsin–Madison (UW–Madison), advised by Professor Andrea Arpaci-Dusseau and Professor Remzi Arpaci-Dusseau. Her research interests are in distributed systems, storage and file systems, and operating systems, with a focus on improving the reliability of distributed systems without compromising their performance.

“I’ve always been fascinated by distributed systems,” she says. “And the reliability and performance challenges are especially interesting at the scale of Facebook with multiple applications and millions of users.” In February, Aishwarya and her team at UW–Madison won a Facebook research award in distributed systems with their proposal entitled “BlockCheck: Examining and improving blockchain reliability.”

Aishwarya learned about the Facebook Fellowship program at OSDI. “One of the main reasons I applied [to the program] was the opportunity to be part of a cohort of peers who are likely to be my future colleagues in academia or industry,” she says. She also appreciates having connections to Fellows from previous years who could provide PhD guidance and practical advice such as what the interview process after graduation is like.

“Winning the Facebook Fellowship also gives me an opportunity to interact with Facebook researchers and engineers. Through these interactions, I hope to learn how the ideas developed as part of my thesis are applicable in the real world.”

Aishwarya’s research has been recognized with a Best Paper award at FAST 2020 and FAST 2018 and a Best Paper award nomination at FAST 2017. She was selected for the Rising Stars in EECS 2018 workshop. Read more on her webpage.

Accelerating deep learning on mobile and IoT devices

Kartik Hegde is a PhD student at the University of Illinois at Urbana-Champaign, advised by Chris Fletcher. He received a bachelor of technology degree from National Institute of Technology Karnataka (NITK), Surathkal, in 2015. His research is focused on building efficient chips to power next-generation AI workloads.

Every day, we interact with smart devices that offer information and convenience through services hosted in the cloud — but this requires a constant internet connection, and response times can be slow. Kartik’s research aims to build more efficient processors that enable more machine learning on the device, faster and more cost effectively.

Before Kartik became a Fellow, he spent some time as an intern at Facebook and experienced what it’s like to conduct research in industry. “At Facebook, we get the best of what academia and industry offer,” he explains. “In academia, you have the independence to pursue your own ideas. In industry, you get the satisfaction of having an impact on people. If your idea is built into something at Facebook, it’s impacting billions of people. It’s the best of both worlds.”

The Fellowship Award includes a paid visit to Facebook HQ in Menlo Park for the annual Fellowship Summit, where Fellows may present their current research in a poster session to discuss their ideas with other Fellows as well as Facebook researchers. “It’s a great privilege where you meet all of the other Fellows and get a chance to network and hear great talks,” says Kartik, who attended the 2019 Fellowship Summit. “I was inspired by conversations I had with other Fellows about new ideas and possibilities for collaboration.”

Although the event won’t be held in person this year, the Fellows will still get a chance to connect with each other remotely. “In addition to the benefit of making new connections, the Fellowship also offers the recognition of winning a competitive fellowship and the independence to work on your own projects,” Kartik says.

To learn more about Kartik and his research, visit his webpage.

Teaching robots to identify objects by combining tactile sensors with computer vision

Maria Bauza Villalonga is a PhD student in robotics at Massachusetts Institute of Technology, working with Professor Alberto Rodriguez. She received bachelor’s degrees in mathematics and in physics from Polytechnic University of Catalonia in Spain. Her research focuses on combining tactile and visual information to enhance robot dexterity during object manipulation.

“In industry automation, many processes involve picking up objects and performing tasks,” Maria explains. “While it’s trivial for a human to identify an object through touch, it requires a lot of intelligence for a robot to pick up an object from a bin and accurately localize it in order to perform the next task. There are many cases where industry cannot yet use robots because of this challenge.”

Maria’s goal is to improve a robot’s ability to accurately manipulate objects and perform tasks by combining computer vision with tactile sensing information. With her MIT-Princeton team, Maria won the Amazon Robotics Challenge stowing task in 2017 and received the Amazon Robotics Best Systems Paper Award in Manipulation in 2018. She explains that through this competition, they learned about the importance of tactile localization for a robot to be able to estimate where an object is.

In her research, the robot uses tactile sensors to scan and reason about the object it is touching and combines that with computer vision to accurately identify and localize objects. She is using tools that Facebook has developed for image recognition and segmentation in her research.

When she finishes her PhD, Maria plans to apply for a faculty position. She enjoys academia because she can think long-term and work on projects that could take years to accomplish, and explore her original ideas with her students. Learn more about Maria’s research on her website.

Advancing sustainable development with just design of human-centered AI and public policy

Neil Gaikwad is a doctoral student in the MIT Media Lab at the Massachusetts Institute of Technology, working with Professor Danielle Wood. He specializes in human-centered machine learning and public policy for sustainable development. Neil develops data-driven human-AI collaboration algorithms and social computing systems, with a focus on increasing fairness and transparency in high-stakes policy decisions.

“Over the centuries, socio-technical systems such as market institutions and humanitarian systems have been instrumental in socioeconomic and environmental sustainability. However, lack of scientific understanding of underlying social and physical processes often leads to prejudiced policy decisions that harm vulnerable communities and hinder sustainable development,” Neil explains. “My research seeks to decode the fundamental nature of these processes on sustainability through the lens of human-centered machine learning, participatory design, and evidence-based policymaking.”

Neil became interested in the Fellowship because of the high-impact research of Facebook Core Data Science (CDS), which aligned with his research agenda. Neil and Facebook CDS have created the first workshop on Humanitarian Mapping at the ACM KDD conference. “The ongoing crisis of the COVID-19 pandemic exemplifies the criticality of the workshop. This initiative will create a framework for researchers to showcase how industry, academics, government research labs, and policymakers can work together to address humanitarian and sustainability challenges,” he says.

Neil is also a recipient of the Karl Taylor Compton Prize (MIT’s highest student award) and is an MIT Arts Scholar. He harnesses contemporary photography to communicate and advocate for science. “It’s hard to contextualize sustainability challenges such as climate change to a general audience merely through scientific models,” he says. “The arts play a vital role in helping people understand scientific underpinnings at subconscious levels and facilitate inclusive dialogues and engagement between diverse communities.”

After graduation, Neil plans to pursue a career in a line of academic research as a professor. “My major goal is to establish an informed research agenda at the boundaries of computational, design, and socioeconomic thinking, with students and global collaborators, to address the pressing sustainable development problems,” he says. For more information about Neil’s work, visit his webpage and arts portfolio.