Kamil Sindi

Kamil Sindi is the chief technology officer of Runway, the company developing generative video and interactive world models. Runway appointed him CTO in February 2026. His work sits at the intersection of AI research and the engineering needed to make generated worlds respond in real time.[2][1]

In a Latent Space interview, Sindi described WorldPrompt, Runway’s way to specify a simulated scene and issue actions as it unfolds. Instead of writing a script for every state of a virtual world, a user can prompt synchronized video and audio in different settings. The feature was part of GWM Worlds 2, a research preview.[1]

Keeping a world running raises a practical systems question: what should the model remember? Sindi said the team must decide what context to retain or discard so long-running generations do not exhaust GPU memory. These choices affect how much of an earlier scene can inform what happens next.[1]

Evaluation gets harder as worlds gain more prompts, characters and scenes. Sindi asks how to determine which interactions caused an outcome. He also points to synthetic data for agents as a possible use for world models, alongside interactive entertainment.[1]