Latent Space: The AI Engineer Podcast
MurmurCast publishes AI-generated summaries of Latent Space: The AI Engineer Podcast’s Podcast episodes — 2 summarized so far, covering Modal's evolution from workflow orchestration to AI infrastructure, Elastic GPU inference and auto-scaling across cloud providers, Agent experience vs. developer experience optimization, Speculative decoding and Dflash for inference optimization, Sandbox primitives for agent deployment and execution, Multi-node training with RDMA networking. Each summary distills the key insights, topics, and takeaways so you can decide what’s worth your time before pressing play.
Why AI Infrastructure must evolve for Agent Experience — Akshat Bubna, Modal CTO
Akshat Bubna, CTO of Modal, discusses how the company evolved from a developer experience platform focused on workflow orchestration into an AI infrastructure provider specializing in elastic GPU inference, agent sandboxes, and specialized compute workloads. Modal's key differentiator is building primitives optimized for bursty, specialized workloads across 17 cloud providers rather than competing with traditional cloud providers on web server infrastructure.
🔬 The Coolest Diffusion Research Isn't in LLMs — Evan Feinberg & Sergey Edunov, Genesis Molecular AI
Evan Feinberg and Sergey Edunov from Genesis Molecular AI discuss how diffusion models have become the superior primitive for protein-ligand structure prediction compared to GANs, enabling sub-angstrom accuracy predictions that are finally useful for drug discovery. They explain their focus on small molecule drug discovery, the importance of synthetic training data from physics simulations, and their shift toward agentic systems while maintaining human-in-the-loop validation.