Blur faces with SageMaker, NVIDIA Triton+Ultra-fast PyTorch data loading from S3
July 18 @ 12:00 pm - 1:00 pm EDT
Talk #0: Introductions and Meetup Announcements By Chris Fregly and Antje Barth
**Talk #1: Blurry faces: Training, Optimizing and Deploying a segmentation model on Amazon SageMaker with NVIDIA TensorRT and NVIDIA Triton**
by Francesco Pochetti
Abstract: Responsible AI is a top concern in the ML world. Can we do anything about it? What if we built a model to protect users’ privacy and automatically blur faces in a photo? Is that even possible? Yes. Not only will we train a neural network from scratch to identify human faces. We’ll also make it blazing fast by deploying it on Amazon SageMaker with NVIDIA TensorRT and NVIDIA Triton. Join us in this end-to-end ML journey from data to production on AWS!
Speaker Bio: Francesco is an ML Engineer working on Computer Vision at Bolt, an Estonian company in the mobility and food delivery business.
He is a chemist turned data scientist, who transitioned from the aerospace sector to analyzing books sales in Amazon Kindle, then building scoring models in the micro-lending business, and now working on identity verification at Bolt. In his spare time, he is an AWS ML Hero and he likes blogging and sharing his experiences with other AI enthusiasts!
**Talk #2: PyTorch’s new native integration with S3 for ultra-fast data loading**
by Joe Evans, Engineering Manager at AWS, and Daiming Yang, Software Engineer at AWS
Zoom link: https://us02web.zoom.us/j/82308186562
O’Reilly Book: https://www.amazon.com/dp/1492079391/
GitHub Repo: https://github.com/data-science-on-aws/