Software Engineer, Compute Infra

About the Role

As a Software Engineer on Compute Infra, you'll build the sandbox infrastructure to give every agent their own VM. These VMs power coding agents, Grok Bot, reinforcement learning, and more. We operate one of the largest set of sandbox clusters in the world, at millions of CPU cores, across third- and first-party datacenters. You’ll own developing a cloud for agents, including the virtualization layer (Firecracker and Cloud Hypervisor microVMs), control plane scaling, and the networking/connectivity layer on top.

Responsibilities

• Rearchitecting the control plane / scheduler to support 1 million concurrent VMs per cluster.
• Building a ublk block device that can dedupe, lazy load, and efficiently snapshot VM data.
• Designing an egress proxy and network isolation to securely filter, monitor, and route agent traffic.
• Designing resilient systems that can operate across a mixture of cloud and on-premise datacenters.

Basic Qualifications

• Bachelor's degree in computer science, engineering, math, or scientific discipline; OR 2+ years of professional experience building software in lieu of a degree

Preferred Skills and Experience

• You have deep experience with building compute orchestration (e.g. Kubernetes, Mesos).
• You have built and operated large fleets of VMs / isolated sandboxes.
• You have experience with the Linux kernel and virtualization technologies.
• You enjoy thinking about infrastructure primitives that future agents will require.