Chris Sosa on Scaling AI Developer Platforms with AMD | AI INFRA SUMMIT 2025

📍 Recorded live at AI INFRA SUMMIT 2025 — Microsoft Silicon Valley Campus What does it really take to scale an AI platform across thousands of GPUs and developers? Chris Sosa, Director of Engineering at AMD, pulls back the curtain in this deeply practical, high-signal keynote from AI INFRA SUMMIT. From ROCm and Radeon to smarter orchestration and multi-tenant optimization, Chris shares how AMD is building for developer-first scale—across code, people, and machines. Whether you're running inference on a single GPU or managing a fleet across Kubernetes, this is your guide to AI infrastructure at scale. 💡 Topics Covered: - AMD’s ROCm stack: drivers, compilers, and open-source evolution - Making Radeon chips dev-friendly (Windows support + legacy reach) - Platform blueprints for multi-user GPU environments - Smarter orchestration + fractionalization for GPU bin-packing - How AMD is building faster repair flows and operator-based recovery 📣 Pre-register for the next summit → https://lu.ma/aiinfra_4 🌐 Learn more at: www.aiinfra.live #AIInfra #ChrisSosa #AMD #ROCm #DeveloperInfra #TokenEconomy #LLMops #AIInfrastructure #Kubernetes #GPUPlatforms