DJ Patil transformed from math dunce to whiz over a single weekend.

“The first math class I failed was in 8^th grade,” he says. “I progressively kept failing my math classes through high school so I barely graduated, then petitioning to get into the next level and then failing that one. I couldn’t get into college so I went to the local junior college.”

To get into the UC system at that time, he had to be ready to take Calculus within one quarter—so he signed up for the course at De Anza College. “I went in there and realized, I don’t know any of this stuff. I had been in the stupid math classes,” he says. “I decided this was ridiculous and I’d like to go to college.”

He went to the Cupertino library, checked out a stack of textbooks, and, over the weekend, taught himself all of high school math. “I remember this moment where my girlfriend at the time was saying, ‘Hey, you’re spending a lot of time on this math class,’ and I said, ‘It’s not like I’m going to be a mathematician or anything,’” he recalls. “Famous last words.”

He aced the Calculus class, petitioned his way into UC San Diego, finished the math curriculum there in three years, and spent his fourth year at UCSD doing research with two professors who suggested that he meet James Yorke, a professor at the University Maryland who coined the term “chaos” in the context of mathematics.

DJ ended up staying there for a Ph.D. in Applied Mathematics. Again, he started at the back of the pack (“I was one of the American-trained students, competing against the Russians, the Israelis, the Koreans,” he says. “I horribly failed my first qualifier, I think I was the second-lowest score and the lowest score was someone who didn’t show up.”). But again, the next time he took it, he was the high scorer.

His graduate research did well and he created a faculty position for himself, and worked on characterizing the complexity of weather and showing that it wasn’t as chaotic as everyone had thought. He’d been running simulations with very large data since he was an undergraduate, one time requisitioning all the computers and bandwidth in the math department every night from 1 a.m. to 8 a.m. for a month to find and process weather data.

He eventually became interested in computational social science and was brought into the U.S. government as an AAAS Science and Technology Policy Fellow to help run some programs in that area. Among other projects, he worked on bioweapons proliferation in Central Asia (and ended up with diplomatic credentials), and co-created the Iraqi Virtual Science Library.

At that point, in 2006, DJ was deciding whether to return to academia or try something different. He had written a paper for the United Nations on using technologies like Skype in places like Iraq, and the Skype team contacted him and asked him to work at eBay. Research funding was tight at that time and he’d always been interested in industry, so he moved back out to the Bay Area.

“When I got to industry, I realized there’s a huge wealth of problems to work on, and they were equally challenging,” he says. “One of the first places I applied data in industry was building new ways of fraud detection systems. We rebuilt the whole frontend security models for eBay and Paypal.”

Meanwhile, he was a member of the board of UC Santa Cruz, along with Reid Hoffman, co-founder of LinkedIn, and the two became friends. “I started giving him advice about how to think about analytics and LinkedIn, and he gave me advice about how to think about Silicon Valley and the industry,” he says. “It was only a matter of time before we said, why aren’t we working together?”

So in 2008, DJ ended up at LinkedIn. “That’s the place where data science finally got formalized,” he says. He and Jeff Hammerbacher, who was at Facebook, came up with the term “data science” for what they did. “We were trying to say, What is the new way you can take data and make it into a user-facing product? That’s really where data science coalesced into something very unique.” He calls it “data jiu-jitsu,” the art of turning data into a product. Now LinkedIn’s data-driven products include “People You May Know,” “Jobs You May Be Interested In,” “Groups You Might Like,” and more.

“At that time in most places, data scientists were relegated to the sidelines,” he says. “We did something totally different, we made it a top-line product team. There’s this general recognition now that if you use data correctly, it’s such an amazing multiplier for the user experience.”

After about three years at LinkedIn as Head of Data Products and Chief Scientist, he had fulfilled his two promises to Reid Hoffman of making data a core component of the company, being done by one of the best data teams in the world. DJ wanted to move on to an earlier-stage company, so he worked for a while at Color, a mobile social application for photos. While he still helps out there, he says, he decided to leave in order to gain more hands-on experience on how to scale an organization.

These days, he’s a Data Scientist in Residence at Greylock Partners, thinking about “the next fun thing to do” and learning about scaling organizations, while simultaneously advising portfolio companies on how to think about data.

As for why he ended up loving math so much, after his initial struggles with it, he says: “Math for me is one of the most powerful set of tools to understand the physical world and make sense of why things happen.” He had simply not realized the practical importance of it when being taught in high school classrooms.