QED has been developing technology to help agricultural development projects efficiently tackle a variety of operational challenges, including survey design, crowdsourcing, field data collection, remote sensing, cloud computing, ML, and map visualization.

We have developed learning technology to attack many industrial problems requiring automated classification and prediction, in spaces such as tourism, credit cards, finance, biology, ag, and cybersecurity. Our training in both “traditional” statistics and modern machine learning often enable the construction of solutions backed by more rigorous justification and theory.

Agricultural, ecological, health, and socioeconomic problems are Big Data problems, but where is all the spatial-temporal data that we need to make accurate predictions and actionable decisions? The need to systematically collect, manage, mine, and visualize this data, particularly in the development world, is more urgent than ever before. SPEED provides a modular and scalable framework to service all of these needs.

QED has developed novel algorithms for tackling a wide variety of challenging and exciting computer vision problems, including biological cell segmentation, 3D imaging of subterranean structures, and the deshredding of shredded documents (DARPA contest). We have also developed open source software in this area, such as template matching in OpenCV.

QED has developed end-to-end NLP data processing workflows for organizing unstructured text, prototypes for automated sentiment analysis and entity extraction of online commentary, and simulated conservations. We have  experience working with most of the major NLP toolkits and databases available today.

Cloud computing is essential to our day-to-day operations for efficiently executing analytics on big data at affordable costs. We are fluent in using IaaS suites from AWS, DigitalOcean, and Rackspace to build scalable systems for data storage, computation, application hosting, and load balancing. QED is also experienced with manipulating data processing frameworks such as Hadoop and Spark, and we have also designed our own customized message-passing alternatives to these tools.

We employ professional software engineering practices, including VCS, CI, buildsystems, project management, code review, and nearly exclusive reliance on open source libraries in a UNIX-based dev environment. We have run git training sessions, contribute to open source development, hackathons, and the upcoming GSOC.