Learn to deploy open-source Generative AI workloads on OpenStack infrastructure in this conference talk featuring Julian Pistorius, Mike Lowe, and Martial Michel. Discover how OpenStack's modular, open design enables organizations to run powerful AI models using existing infrastructure while maintaining data sovereignty, cost control, and flexibility. Explore practical deployment strategies for open-source GenAI software, including Large Language Models (LLMs) and Stable Diffusion, on GPU-enabled OpenStack environments using containerized services and subnet-routable floating IPs. Examine real-world examples of building secure, scalable, self-hosted GenAI services for internal applications and research purposes. Understand how deployed models can support advanced use cases such as local embeddings for RAG (Retrieval-Augmented Generation) pipelines, semantic search capabilities, agent-based automation systems, and LoRA-based fine-tuning techniques for domain-specific performance optimization.