A PyTorch Approach to ML Infrastructure

Explore a 31-minute conference talk on a novel approach to ML infrastructure presented by Caroline Chen and Rohin Bhasin, Software Engineers at Runhouse. Delve into the limitations of traditional model and pipeline portability methods for scaling ML across organizations. Discover a proposed alternative that focuses on leaving ML methods in place while making them instantly sharable and accessible from anywhere. Learn about open source PyTorch-like APIs designed to create and share production-grade, scalable shared apps and services. Gain insights into how this approach aims to address the fragmentation in enterprise ML and provide agnosticism to underlying infrastructure and existing ML tooling.