Member of Technical Staff, Sandbox Infrastructure

About the role

We are building a serverless GPU container service that runs untrusted kernels safely across NVIDIA, AMD, TPU and Trainium. The Member of Technical Staff will extend our sandboxing stack, add new vendor support, and ensure high‑fidelity measurement for our compiler’s search process.

Key responsibilities

• Extend the sandboxing stack to additional accelerators and vendors.
• Develop and maintain GPU virtualization below the runtime, including driver‑level work with gVisor.
• Implement preemption‑aware scheduling, checkpointing, and rescheduling for spot capacity.
• Support multi‑GPU and multi‑node sandboxes, handling NVLink, NVSwitch, InfiniBand and RoCE interconnects.
• Own end‑to‑end live migration with socket‑preserving relocation.
• Guarantee measurement and profiling fidelity while preserving isolation.

Required profile

• Strong low‑level systems engineering experience (Linux kernel, containers, namespaces, cgroups, syscall interception or hypervisors).
• Hands‑on GPU systems expertise (drivers, runtimes, scheduling on accelerator fleets).
• Experience with distributed‑system failure modes such as preemption, checkpoint/restore, and state preservation.
• Proficiency in Go, C/C++ or Rust and a bias toward building custom solutions when vendor support is lacking.

Required skills

• Linux kernel internals
• Containers, namespaces, cgroups
• GPU drivers and runtimes
• Go, C/C++, Rust
• gVisor, Firecracker, Kata, QEMU/KVM
• CRIU live migration
• NCCL/RCCL, RDMA, InfiniBand, RoCE

What we offer

• Base salary range $275,000‑$315,000 per year.
• Meaningful equity participation.
• Comprehensive benefits package.
• Relocation assistance and an onsite San Francisco office.