Type Private Website www.mapd.com Founded 2013 | Headquarters San Francisco Founder Todd Mostak | |
 | ||
Industry Software for data analysis Products Database Analytics and Visualization Profiles | ||
MapD Technologies builds database and visualization applications that take advantage of the parallel processing power of graphics processing units (GPUs). Customers include the US government.
Contents
History
MapD Technologies was founded by Todd Mostak and Tom Graham in Boston in 2013. The idea came out of Mostak’s graduate work at Harvard’s center for Middle Eastern studies, where Mostak studied from 2010 to 2012. Mostak was studying the Arab Spring and in particular, the role that social media played in that movement. Unable to analyze the millions of tweets, Mostak turned to GPUs to visualize the data. While at Harvard, Mostak took classes at MIT’s Computer Science and Artificial Intelligence Laboratories (CSAIL) taught by professors Michael Stonebraker and Sam Madden.
In 2013, Mostak partnered with Harvard’s Center for Geographic Analysis to build a tweet map demonstration. Mostak then joined Harvard’s Kennedy School as a research fellow on Quantitative Political Science on Middle Eastern issues. Later that year, Mostak joined CSAIL as a research fellow, officially incorporating that fall.
In January 2014, Mostak left MIT to pursue MapD full time. Two months later, MapD won a prize from GPU maker Nvidia. That fall, Mostak relocated the company from Boston to San Francisco.
In the spring of 2015, the company closed a seed round that included investments from Google Ventures (now GV), Nvidia and Vanedge Capital.
On March 30, 2016, the company announced general availability of its database and visualization products as well as the company’s $10 million investment led by Vanedge Capital with additional participation from Verizon Ventures and Nvidia.
Software
The company builds and sells the MapD database management system product and a visualization layer.
The MapD database is specifically designed for GPU environments and takes advantage of both the memory bandwidth and the massive parallelism available on that hardware. By tuning the database for these hardware capabilities, MapD can execute queries at 100x the speed of traditional CPU databases.
Furthermore, by using the video rendering capabilities for which GPUs were initially designed, MapD has built a visualization layer called Immerse that is capable of rendering large datasets (billions of rows of data) with millisecond lag.