Session abstract
What you’ll learn
Open-source video generation has crossed a threshold. The base models are capable, and the tooling is finally approachable. This session walks through the full stack. We'll cover how to curate and structure training data for video fine-tuning, how to run a LoRA or full fine-tune on a modern DiT-based video model without burning your compute budget, and how to get that model wrapped, versioned, and deployed to a serverless GPU provider so it cold-starts in seconds.