Sridhar Ramaswamy PhD ‘95 is one of these.
An honored guest, he’s returning to Brown to deliver a lecture (“F1: A Distributed Database That Scales”) that will inaugurate Brown Computer Science’s new Information Technology Leaders Lecture Series. Brown CS warmly welcomes Sridhar at a time of exciting growth for the field, the university, and the department.
Arriving at Brown CS in the early 1990’s, Sridhar found an environment that was “wildly different” from his previous schooling at Indian Institute of Technology (IIT) Madras, where the approach to learning was highly structured: “Brown offered a number of areas that I was interested in at the time, like databases and graphics, and the enormous amount of intellectual freedom to do what you wanted.”
Like many others, he gladly seized the many opportunities of the open curriculum. “I used it,” he says, “to learn many wonderful things. I took lots and lots of courses on literature and music. Brown was amazing in that respect. It really opened the rest of the world to me, things that I didn’t get from a super-technical education.”
Sridhar’s observations about the size of an earlier Brown CS (“It was small at the time, about fifteen professors. Whatever area you wanted to study, you had one choice of advisor.”) anticipate the growth that occurred in subsequent years as the Department of Computer Science doubled the size of its faculty. Small or not, Sridhar remembers the years fondly, even recalling the course number of a legendary class: “I loved Stan Zdonik and Tom Doeppner, how open they were. You could talk to any professor, any of the grad students. It was quite cool. I still remember CS 169!”
A single sentence from a letter that Sridhar sent to Conduit in 1999 nicely frames the two sides of the career that immediately followed his graduation: “I am leaving the comfortable and laid-back environment of Bell Labs for the thriving madness of Silicon Valley.”
The seeds of this desire for “thriving madness” were sown with his early research work: “At Brown, I discovered all these amazing problems to be solved, and wonderful, inspiring people like Paris Kanellakis, who was my advisor. That desire to solve problems built on itself and kept growing. I’m a bit of a stubborn person; I didn’t want to start something and leave it half done.”
“Part of Paris’s advice to me,” Sridhar says, “was that I needed to become more of a systems person, which was what I had done as an undergraduate at IIT with things like graphics device drivers. I started doing more database systems research, first at Bellcore, then Bell Labs.” This blend of theory and practice continued with Ramaswamy’s work on a revolutionary system called AQUA (Approximate QUery Answering) that was aimed at providing provable guarantees for fast answers on massive datasets. “It was a collaboration with a wonderful group of people,” he remembers.
But this was the Internet Age. As Sridhar explains, “My thinking was that I wanted to change direction, and if I was really going to do so, I had to make a drastic change. I decided that I’d head out to the Valley to try my hand at software engineering.” After four years as the Director of Engineering at a startup called E.piphany, a slightly larger company beckoned in 2003. It was Google. “Well, it was obviously a place that was full of opportunities, and I thought it was humongous when I first joined,” he laughs. “I think there were all of three hundred engineers!”
Sridhar pauses for a moment, mentally reviewing the past decade, which has brought him to the position of Google’s Senior Vice President of Advertising and Commerce, where he oversees the design, innovation, and engineering of the company’s advertising and commerce products. Leader of the engineering teams that helped define the vision and direction of AdWords, he now also leads Google’s efforts in Display advertising, Analytics, Shopping, and Payments. Additionally, Sridhar is part of a group of senior executives who report directly to CEO Larry Page.
“Gosh, I’ve had a wonderful time at Google.”
At length, he explains that much of his success comes from “being part of an amazing team” in an equally incredible environment. “None of this,” he insists, “is mine alone, but one thing I’m very proud of is the Advertising teams’ emphasis on building amazing infrastructure. We believe that creative systems engineering goes a long way toward solving problems that are otherwise considered unsolvable.”
Interestingly for someone who transitioned from researcher to software engineer, some of Ramaswamy’s favorite achievements are process-related, systems-oriented, even theoretical. “We’ve developed an entire science,” he says, “around how we think about experiments and how we deploy them. All the changes on the site, big or small, can have a huge impact in areas such as monetization. To be on a team that helped develop a whole framework to ask how one considers change, how one develops experiments and conducts them in massive numbers while making sure they don’t interact with each other, that’s a signature accomplishment for me. It’s a cool thing.”
Sridhar also mentions his love of auctions, saying that he’s been privileged to be a part of the “considerable amount of innovation” in this area. One of the greatest challenges for Google or any of its competitors, he says, is striking a balance between the long-term value of auctions and their current value while trying to optimize for the long term.
But the accomplishment that Ramaswamy actually mentions first is the one that will serve as the subject of his inaugural Information Technology Leaders lecture. F1 (the name is derived from the nomenclature in Mendel’s famous genetic experiments, not the type of racecar) is a hybrid between an object store and a relational database. “We wanted a system,” Sridhar says, “that could be what we call planet-scale but also answer very difficult queries for the purpose of, say, building reports. We wanted both things, and we did it. It’s in in production now. The conventional wisdom was that you could build scalable object stores or relational databases, but not both together.”
Circling backward in the conversation to give a precise definition of what “planet-scale” might refer to, there’s a soft but unmistakable satisfaction in his voice. “We mean a system that will keep working beautifully even if California becomes an island.”
A Very Different Future
“There is a world we’re already living in that I think will become even bigger and more important.” Asked about Brown and the digital society of the future, Sridhar’s answer is immediate. “That’s the world of cloud computing and amazing devices and interfaces.”
He begins with a bit of history: “When the first Internet companies started, people were literally running services from their garages. Then computers got better, people put them into racks, but there was always this sense that you had to own big machines, big data centers. In my mind, what’s been remarkable about the last few years, as evinced by many companies like Netflix and Dropbox, is that cloud computing is such a commodity and can run at such a vast scale.”
The opportunities for the next wave of computer scientists and the next generation in general are omnipresent. “Four people with exactly four laptops can build global-scale applications,” says Sridhar. “That was unimaginable ten years ago. Things get more and more refined, and people have mobile phones in their hands, these wonderful interfaces. Behind them, a back-end is running at an incredible scale. It’s pretty amazing. People don’t quite appreciate it, but it’s breathtaking.”
He’s also very excited about the connected Internet (webcams, thermostats, and even house locks that can be operated remotely from a phone) and the “very different future” it foretells. On a more narrow technical front, advances in machine learning and artificial intelligence have tremendous untapped potential.
“People don’t even realize,” says Sridhar, “what we can do with simple models and massive amounts of data. You might think that spelling correction uses fancy logic about sentence structure or phonemes, but it’s purely a statistical algorithm based on a dataset of errors and corrections. The whole field of deep learning is young, and it’ll solve problems that look quite beyond our capacity right now.”
A statistical algorithm to analyze Sridhar Ramaswamy in conversation would provide equally interesting insights. Three times in a half-hour, he’s used the term “conventional wisdom” to point out the conclusions that many people readily jump to, the dead end offered by cynicism, a failure of vision. Inspired by mentors such as Paris Kanellakis, Tom Doeppner, and Stan Zdonik, drawn to collaboration and to lead teams of people that he respects so highly, one challenge has echoed throughout Sridhar’s responses. “It’s about solving,” as he puts it, “the unsolvable.”
It’s a system that can operate if California becomes an island.
It’s four people with four computers making global-scale applications.
It’s a world in which anything is possible.
“Computer science is what’s really exciting,” says Sridhar. “I say to my children, you can be a physicist. It’s an honorable profession. But every smart person for the past five hundred years has been trying to solve the hardest problems of physics. What excites me is that computing itself is very young. It’s only begun to touch numerous aspects of our life. I can point kids to so many different areas and tell them that there are thousands of possible futures there, just waiting!”
At a time when the field’s prospects and Brown University’s have seldom looked brighter, Brown CS looks forward to Sridhar Ramaswamy’s inaugural lecture, his unconventional wisdom, and the opportunities that those futures offer.
Brown CS cordially invites you to attend Sridhar Ramaswamy’s lecture (“F1: A Distributed Database That Scales”), which will take place on September 18 at 4pm in CIT 368 at 115 Waterman Street. Everyone’s welcome, and remote attendees can watch the live stream by clicking here. Shortly after the lecture, we'll make a recording available here.