When Ph.D. data scientist John Kaufhold left his job at the National Institutes of Health in 2013 to start a company of his own, his motivation was fairly simple: He wanted to do interesting, innovative work in a collaborative team with really smart people.
At that time, the field of deep learning was taking off. Deep learning is a form of machine learning that uses a computing model inspired by the structure of the brain. It requires less human supervision and is used widely in artificial intelligence. John had previous experience in deep learning, so he focused his initial efforts on developing new techniques in this area. He called his company Deep Learning Analytics.
A year into his venture, John invited his friend Sue Rego, now the company's chief operating officer, to run the business. Sue had managed three small technology companies in the past. Soon, she was on board, taking responsibility for everything from employee benefits to accounting to managing the company's cost proposals that are critical when bidding on work for the federal government.
Next, John recruited Henry Chen, a software engineer he'd worked with in the past. Henry set up the company's collaborative development infrastructure, helped shape the company's culture and aided John in bringing on a team of skilled data scientists and engineers.
Today, Deep Learning Analytics employs 12 people who work at the Rosslyn campus of Eastern Foundry. While Sue runs the business, Henry and John manage technical execution and collaboration. Taking on innovative assignments for DARPA as well as other data science projects have enabled Deep Learning Analytics to grow.
The company's most public accomplishment thus far has been winning TRACE, a $6 million DARPA contract. Now in the second year of this 42-month program, they are exploring how deep learning techniques can improve the military's analysis of radar images, enhancing their ability to identify targets rapidly and accurately. The story of how they won this contract in the face of intense competition was detailed in an August 2016 Washington Post article. "It really is a David and Goliath story," Sue says.
Deep Learning Analytics has also won Arlington's Fast Four Competition two years in a row. The honor recognizes them as one of the region's fastest growing startups on the basis of year-over-year revenue growth.
"There are so many ways we got lucky in this whole risky startup world — we were able to attract the best data scientists in the D.C. metro region to work on a really, really interesting problem right at the time when the market for this kind of disruptive work in AI and deep learning is really taking off," Rego says. "And we built this really collaborative culture that grows the whole team technically every day.”
Adds John: "We wouldn’t have been able to do any of the work we do if we didn’t have Sue’s business experience to navigate the complexities of government contracting. It’s the best professional experience of my career and I’m thankful for every day I get to work with this amazing team.”
So what's next for Deep Learning Analytics?
Sue says she's optimistic.
“Deep learning is really hot right now, so we’re still growing by leaps and bounds," she says. "But with the surging interest in deep learning and artificial intelligence also comes some ‘irrational exuberance’ we’re trying to steer clear of as a business. Probably the biggest challenge we’re facing is trying to maintain that balance between holding on to our technical culture at the same time we grow at a healthy clip. So far, so good.”
Henry adds that the whole team continues to outperform expectations on execution, making management of such a complex technical team that much easier. John agrees, adding, “We’re really lucky to be in this space right now. Deep learning is to 2015 what the internet was to 1995. The companies that successfully made it past 2000 were the ones that were solving real business problems and didn’t get taken in by the dotcom hype. We believe our focus on execution and practical software engineering solutions to manage technical debt is what really differentiates us from competitors.”
Learn more: www.deeplearninganalytics.com