Projects at Microsoft

  • Land cover mapping with Esri and the Chesapeake Conservancy
    • Released a tutorial demonstrating how our semantic segmentation model was developed on a single GPU VM and scaled up to a GPU cluster on the Azure cloud (blog post, repository)
    • Presented our work at the CVPR DeepGlobe workshop (2018), the Esri Dev Summit (2018), the Visipedia Workshop (2018), the AI for Earth Summit (2018), Microsoft Ignite (2017), and Microsoft-internal conferences and events
    • Enabled distributed training and real-time inference using two early-stage Azure services (Azure Machine Learning Services and Azure Batch AI), providing feedback during product development
    • Assisted with nationwide land cover mapping and design of a feedback collection system
  • Object and structure detection from overhead imagery for the xView Challenge
  • Crop performance prediction using genomic and environmental data
  • Project Brainwave (FPGA web service) stress test
  • Inference with DNNs on Spark
  • Recommendation systems on the Azure cloud
  • Retail price optimization guide
  • Python port of the Predictive Maintenance Modeling Guide