Session: Open Source-powered and Developer-first AI Stack
Enterprise technology leaders are turning down many worthy AI projects due to the punishing cost of hyperscaler AI tokens. Massive investments in GPUs from a single vendor end up underused due to poor software solutions. This AI Infrastructure was built with graphics in mind rather than the operations that commonly make up AI training and inference operations. And the closed ecosystem operates as a black box, limiting engineers’ abilities to optimize performance and fix issues.
This is not an inevitability and we, as an industry, are working on increasing average utilization and increasing diversity of accelerators. The choice, control, and affordability that Open Source and Open Standards have brought to traditional software development the past 40 years have now finally come to AI.
Yann Léger, Co-Founder & CEO of Koyeb, a next-generation serverless cloud for developers, and Greg Wallace, Director of Partnerships at edge infrastructure provider NetActuate, will detail and demo how they deployed novel AI accelerators into the cloud, allowing AI developers from all parts of the stack to access GPU alternatives in seconds. Learn about how they overcame the challenges of deploying novel accelerators in a traditional server environment to put Tenstorrent WormholeTM and BlackholeTM technology in the hands of kernel developers, machine learning engineers, and AI technologists. Discover how you can try novel AI accelerators today and contribute to their optimization.
This session will be recorded
