Shivaram Venkataraman joins Department of Computer Sciences, researching system design in the age of smart devices

Shivaram Venkataraman, who joins the Operating Systems Research Group in the Department of Computer Sciences, is interested in building systems for machine learning. “This involves interacting with machine learning and application-level users to understand what kind of algorithms are they executing, what kind of applications they want to run, and what data properties they have,” says Shivaram. “My dissertation was based on this kind of research, and I continued working on projects related to that during my post-doc, specifically looking at how to improve the efficiency of a large cluster of machines that are all executing machine-learning workloads. How do we share the cluster effectively across a large number of these users? And how do we avoid failures—build resilient systems?”

Shivaram is also interested in looking at Edge computing devices. He says, “All of us carry smartphones, but there are now smart tvs, smart refrigerators, etc. One way to avoid being spied on all the time is to do all the computation within your house—if the data never leaves your house, it’s more secure. There are a number of challenges in doing this, and I’m excited to look at how to design systems to accomplish this.”

Shivaram comes to UW-Madison from a post-doc in the Systems Research Group at Microsoft in Redmond, WA. He completed his PhD at UC-Berkeley, his MS at the University of Illinois Urbana-Champaign, and he started working at Google in Bangalore after earning his undergraduate degree. “One day you’re doing some small assignments in your dorm, and the next you’re at Google and can build an entire web search engine on your machine if you want to,” says Shivaram. “It was a huge jump! I had a steep learning curve, but it triggered my interest in large-scale data processing systems and how they’re built. I was working more on the application end, social networking products, for example, that were built on top of large-scale data processing systems. Looking at the design and implementation and how they were able to scale it, what sort of problems they ran into, kindled my interest in studying more about this, and I decided to go to graduate school.”

Graduate school led to Microsoft which led to UW-Madison. “The Computer Sciences Department at UW is in the top three in the world in systems and databases,” says Shivaram. “Walk down the seventh floor [where faculty offices are located] —it’s unbelievable the set of people who are there. They’re the top of their field and also a really helpful and nice group of people.” Shivaram likes to work collaboratively, and he finds this is welcomed in the UW CS department and across campus. “There’s a general acceptance and encouragement of collaboration, even across departments—I value that a lot,” he says. “The times I’ve learned a lot have always been from working with people whose expertise is different from mine. They bring things to the table, I bring things to the table, and we get out of it more than the sum.”

This semester Shivaram is teaching the graduate course on big data systems. “I’ve spent time designing new courses for data science, and the traditional method of lecturing to students and reading a textbook doesn’t fit very well for that kind of class,” he says. “We tried combining lab and lecture sessions, going back and forth between lecture and lab work, so students were able to act on what they were learning right away. This can be challenging with regards to space and number of instructors to make that work, but it’s something I’m interested in trying at UW-Madison, as well.” Shivaram would also like to enable and encourage more cross-student interaction in his classes. “It would be a lot more powerful to bring this into the classroom.” And he thinks this is a good time to figure out these new models, when the need for CS classes is so large.

Shivaram is happy to be in Madison, in part because it’s a medium-sized city. “In bigger cities like Seattle and in the Bay Area, you can’t get anywhere,” he says. “In Madison, you can get almost everywhere pretty easily.” As a hiker and runner, he’s enjoying being able to get around town, as well as how tree-filled the city is, and is looking forward to cultural events in Madison. Up next, he’ll explore the area outside Madison and tackle winter activities

For more about Shivaram, read his interview with L&S here.