Deepwave Digital at GTC San Jose

If you plan to come to NVIDIA’s GPU Technology Conference in San Jose, CA from March 17-21, come by and see us. The Deepwave team will be in the vendor exhibit area March 19-21. Stop by if you want to chat about the AIR-T, deep learning, or just want to check out the product before yours ships. There will also be a number of technical presentations focused on deep learning applications for radio frequency systems, including a talk by Deepwave.

Manufacturing Has Begun!

We are excited to announce that we are kicking off the manufacturing run of the production model AIR-T this week! Bringing the first deep learning-centric software defined radio (SDR) from an idea to a commercial product has been a long and exciting engineering adventure. The AIR-T has a 12-layer PCB with over 1,300 components. On top of all that, we we want to supply our customers with a familiar and functional software interface for command, control, and operation of the platform. While we hope to get the AIR-T to our first backers sooner, our conservative estimate for final delivery is March 31. We would like to thank our supporters for their patience, generosity, and support.

Photographs of AIR-T

Here are some photographs that we took of the final production model of the AIR-T.

Much more to come, so stay tuned!

More Information on GR-Wavelearner Software

We have put together a presentation explaining how to use GR-Wavelearner to deploy neural networks on the AIR-T, which consists of three steps:

1. Train - Using TensoFlow, MATLAB, Keras, Pytorch, or your favorite deep learning framework
2. Optimize - Take your trained neural network and optimize it using TensorRT
3. Deploy - Using GR-Wavelearner, GNU Radio, and the AIR-T

Check out the slides here for more details.