Meet Daniel Bichuetti - Speaker at KubeCon + CloudNativeCon North America

Why watch

This session explores building a Kubernetes architecture capable of fine-tuning a domain-specific vision language model across 252 GPUs. Daniel Bichuetti covers critical infrastructure topics including topology-aware scheduling to optimize GPU placement, RDMA networking for low-latency inter-node communication, large-scale data pipelines for ML training, and failure recovery strategies. The talk is designed for ML platform engineers, DevOps professionals, and infrastructure architects working on large-scale distributed ML systems.