Accelerating Multimodal AI - From Tabular ML to Generative Video Systems

Explore the challenges and opportunities of multimodal AI in this 55-minute podcast episode featuring Ethan Rosenthal, Member of Technical Staff at Runway. Dive into the complexities of managing and accelerating multimodal AI systems, from data management to efficient inference. Learn about the similarities and differences between tabular machine learning, large language models, and generative video systems. Discover effective setups and tools for supporting both research and productionization processes in the rapidly evolving field of AI. Gain insights into topics such as multimodal feature stores, large-scale distributed training, and the emerging Generative DevOps movement. Understand the challenges of bridging the gap between researchers and engineers in AI development and explore strategies for structuring teams to maximize efficiency in multimodal AI projects.

[] Ethan's preferred coffee
[] Takeaways
[] Falling into LLMs
[] Advanced AI Tech Capabilities
[] AI-powered video editing tool
[] Transition to AI: Diffusion Models
[] Multimodal Feature Store breakdown
[] Multimodal Feature Stores Evolution
[] Benefits of Multimodal Feature Store
[] Centralized Training Data Repository
[] Large-scale distributed training
[32:37 - ] AWS Ad
[] Dealing with researchers on productionizing
[] Infrastructure for Researchers and Engineers
[] Generative DevOps movement
[] Structuring teams
[] Multimodal Feature Stores Efficiency
[] Wrap up