Your partner in
physics-based synthetic data generation

AI is just software that uses data instead of code. We provide a Common Application Framework to produce physics-based synthetic datasets for AI training and validation.

Get in Touch
Synthetic Aperture Radar (SAR) Image Generation

Common Application Framework for Synthetic Data

Nathan Kundtz, CEO, April 12, 2021 10:00am PDT (UTC-7)

It’s not just about having “enough” data.
It’s about control within a Common Framework.

  • Robust physics libraries
  • Automated scene generation and world building
  • Managed high-performance compute
  • Both web and API-based tool access
  • Cloud-hosted data or local downloads
  • Integrated model generation for QA/QC
  • Closed loop between data generation and model outcomes
  • Enables rapid iterations and collaborative workflow
  • A tool for comparing, labeling, cleansing your data
  • Powerful modeling, measuring, and data depiction tools
Generate tens of thousands of images in a managed HPC environment
Procedural world generation and digital twin scraping/imports​

Procedural modeling and geometry generation tools are built in

In our experience, lack of diversity in the dataset is the root cause of most synthetic data failures. To address this issue, we incorporate procedural models over static ones whenever possible:

  • Landscapes, Vegetation, Buildings, Oceans, and Cities
  • PLUS a large library of internal and/or user uploaded models

Globally accessible and ready to scale

We provide an easy-to-use graphical interface for dataset generation AND a complete set of APIs for access wherever you need it.

Data can be downloaded locally or used with cloud-based pipelines (including directly to your AWS S3 bucket) keeping data residency near a global set of analytics tools.

Our architecture is cloud native; meaning almost instantly scalable compute environments are at your fingertips for both dataset generation as well as training and AI deployment.

Node-based data programming for automated creation of a synthetic ATR library
2 sensors, 1 procedurally-generated scene

Flexible and Powerful

Our shared simulation domains allow us to produce cross-domain, integrated, multi-sensor simulations to support robust data-fusion models from a single simulation enivronment.

All of our rendering engines are GPU-accelerated (including our full-wave Synthetic Aperture Radar engine!) meaning you'll have access to massive compute available via GPUs.

If we don’t have something you need in our tools library, we can build it for you or give you API hooks into the simulation environment.



NEW – Get an overview of the Common Application Framework, including a live demo of our web-based user interface.

AI INTRO – Learn more about needs and challenges in delivering prove-able AI, and the benefits of a data engineering approach.

INTERVIEW – Catch Nathan Kundtz, Founder/CEO, discussing synthetic data generation with Adam Simmons of Project Geospacial.

Get in Touch

Drop us a line, ask a question, or request a demo or login.
We’re happy to help in any way.