NVIDIA & Marlowe: Scaling Code Agents for 100M+ LOC

Abstract

This talk will cover how to use NVIDIA GPUs and GPU-accelerated techniques to efficiently index, update and search massive code-bases with over 100M lines of code. Building an index slashes tokens by nearly 70% without compromising speed or accuracy in vibe coding.

Special Speakers | Rohan & James

Rohan is a Senior Solutions Architect at NVIDIA, working with enterprise partners on Generative AI adoption. In his previous role, he worked on the autonomous vehicles team at NVIDIA and prior to that, founded a startup working on advanced driver assistance systems (ADAS) for trucking fleets. Rohan earned a Master's degree in Robotic Systems Development from Carnegie Mellon University.

James is a Senior Solutions Architect on the NVIDIA Media and Entertainment SA team focused on the intersection of AI and video games. James earned his Master's degree in Data Analytics from Pennsylvania State University. Outside of work, he is a passionate snowboarder and gamer.