OmniSci’s GPU system yields rapid insights from vast datasets
Around the time of the 2011 Arab Spring, OmniSci (Booth 1526) CEO Todd Mostak learned what many others were beginning to grasp, that social media had intelligence value. But how much value was difficult to ascertain given the sheer volume of the open-source data.
Mostak had just returned to the U.S. after two years of teaching English and learning Arabic in Egypt and Syria. While taking computer science courses at MIT, he tweeted questions to assess activity in the region where he had lived. Tweets flowed back in astounding numbers, some in answer to his queries, others just offering time-stamped and geo-located reports to the world from the region. But clarity among the massive numbers was elusive “because of the lack of speed of the hardware and software,” Mostak said.
He found a computational solution in video game cards he incorporated in graphics processing units (GPUs) to replace traditional, slower central processing units (CPUs). From that Mostak created San Francisco-based MapD in 2013.
MapD rebranded as OmniSci in 2018, when Mostak realized the company’s GPU system had sent it well beyond mapping capabilities. He claims the GPU system is 75 to 3,500 times faster than a CPU database and has many useful applications in addition to social media analysis.
At GEOINT 2019, OmniSci is showing an aeronautical display that meshes telemetry data points with real-time radar to determine how planes navigate.
As a young entrepreneur, Mostak first attended GEOINT 2014 to learn. Now, he’s attending GEOINT 2019 as a new USGIF member and plans to teach as well as learn.
Seeking expanded agility, collaboration, and innovation, the NGA reimagines its acquisition strategy.