Date: Aug 24, 2016 Author: Dan Beyers Source: Washington Post (
click here to go to the source)
John Kaufhold finished graduate school in 2000, and then began working as an industrial scientist, first at GE's Global Research Center in New York, then at SAIC as a technical fellow, and then as a staff scientist at National Institutes of Health. In those roles at large organizations, Kaufhold led many teams solving various image analysis and machine-learning problems, writing proposals to the government, publishing research and patenting inventions. During this 12-year period, Kaufhold made contributions in mammography capabilities, computer-assisted detection on medical images, and modeling cerebral blood flow in mice, among others. In his last formal role at a large organization, Kaufhold investigated deep learning, a branch of artificial intelligence technology that uses algorithms to spot patterns in digital data, much like the human brain might.
Kaufhold left his position at NIH in 2013 to start his company, Deep Learning Analytics. At that time, Kaufhold sought advice from SCORE mentor Gerry Sanz, who had previously run a small federal contracting firm. Over the next year Sanz guided Kaufhold through the maze of small business start-up processes including getting an office location, a business bank account, incorporating with the state corporation commission, getting insurance, and setting up an accounting system.
Kaufhold understood very early on that deep learning techniques could improve the analysis of radar images, and he reached out to a program manager at a government research agency to discuss possible applications. Eventually the government asked Kaufhold for a proposal to conduct a small feasibility study. Kaufhold teamed with another small company and in a few months delivered promising results. Kaufhold's study led the government to release what is known as a broad area announcement inviting companies to expand on the feasibility study.
Kaufhold knew from experience that winning the bid would be difficult. The accounting and compliance requirements alone would be challenging for his start-up, not to mention the likelihood that several teams of large companies with deep pockets would submit formidable, thoroughly vetted proposals. It seemed to Kaufhold that he had become a victim of his own success. His feasibility study results had created a market for the follow-on opportunity that seemed out of his reach. Indeed, representatives from many large company's came to an agency "industry day" to learn about the opportunity and size up their competition.
Kaufhold understood that even though he successfully executed the initial study, innovation is not enough to succeed in government contracting. Deep Learning Analytics had very little time (60 days) to ramp up. His company had no employees, no revenue, no line of credit, no subcontracting expertise, no compliance structures (not even a time charging system) and very little past performance to support a winning proposal. Kaufhold was told by just about every advisor he asked that it would be impossible to submit a compliant proposal -- let alone a winning one, but still he pressed on.
The first strategic decision was to split the proposal problem in two. Kaufhold focused on the technical approach and recruited as his chief operating officer a trusted friend, Sue Rego, who had already run two small businesses to take full responsibility for the cost proposal. Kaufhold also turned to his SCORE mentor, Sanz, for guidance on how his new company could navigate the complex process of preparing a proposal response.
Deep Learning Analytics grappled with the difficult choice of whether it should subcontract to a prime or try to prime the contract on its own. Kaufhold had to assess the trade-offs. As prime, Deep Learning Analytics could retain more control over the contract, but the probability of winning was lower due to it inexperience and status as a start-up. Subcontracting carried a potentially higher chance of winning with a more established prime, but may not be the best way to build a company for the long term.
With Rego's encouragement to take the risk, Kaufhold decided to bid as a prime contractor, reasoning if he and his new partner lost, they would at least do so knowing what was at stake.
They then went to work putting together a "projected company" capable of doing the work, one that would require multiple subcontractors and about 10 employees, several with hard-to-find data science qualifications. Rego collected required salary justifications and overall business costs for the team, and she and Kaufhold and worked cost proposal spreadsheets together over multiple weekly sessions in Sanz's SCORE office and over email. Sanz walked Rego and Kaufhold through the nuances of how to calculate "multipliers" from "indirect" costs, helping sort out what fringe, administrative and overhead costs would look like.
In 2015, Deep Learning Analytics and their three larger subcontractors learned they had won the award through a fully open competition, besting the likes of Boeing, Raytheon, Northrop Grumman and Lockheed Martin. Their projected company soon became a real one, recognized as one of the four fastest growing companies in Arlington later that year. Nine months into their contract, a Deep Learning Analytics' prototype was featured at the Defense Advanced Research Projects Agency' (DARPA) Demo Day at the Pentagon.
"I had never considered myself an 'entrepreneur' before," Kaufhold said. "Without SCORE, I certainly wouldn't have been positioned to navigate this process. And without someone like Sue handling the business, I also know I wouldn't have been able to focus on the innovative, technical parts of the work--the stuff the government really needs."
SCORE is a nonprofit association dedicated to entrepreneur education. Looking for some advice on a new business, or need help fixing an existing one? The Greater Washington DC Chapter provides confidential counseling and mentoring from more than 50 executives across the region. Contact us at email@example.com or request a mentor at www.washingtondc.score.org.