ONNX and ONNX Runtime with Microsoft's Vinitra Swamy and Pranav Sharma_1

Microsoft hosted an AI enthusiast’s meetup group in San Francisco in November 2019 focused on accelerating and optimizing machine learning models with ONNX and ONNX Runtime. Microsoft’s Pranav Sharma, Principal Software Engineer and Vinitra Swamy, Software Engineer led the discussion.

First up, Vinitra offers an in-depth introduction to ONNX, including explaining how ONNX is useful to hardware vendors, service/framework authors, data scientists, and machine learning engineers alike. Vinitra overviews how the ONNX technical design helps Microsoft overcome deployment issues at large scales in their products.

Next, Pranav dives deeper into ONNX Runtime, the high-performance inference engine for ONNX models, which is completely open-source, cross-platform, and developed on GitHub. He discusses the design principals and considerations when developing ONNX Runtime including backwards compatibility and high performance, among others. Pranav also provides a complete overview of the ONNX Runtime Architecture.

Finally, Pranav and Vinitra demo ONNX and include object detection models that they’ve been using.

Microsoft Bay Area Meetups bring technical groups together to share insights from Microsoft teams based in the region and connect with a community pushing forward innovation.

Check us out on:

#MSFTBayAreaMeetups