S

San Francisco Compute Company Tech Stack

Real-time trading platform for GPU compute capacity

Technology, Information and Internet San Francisco, California 11–50 employees Privately Held

SF Compute operates a spot market for GPU compute, modeled after commodity exchanges. The stack—Kubernetes, SLURM, Rust, Go, vLLM, and deep networking primitives (VXLAN, RoCEv2, InfiniBand)—reflects infrastructure-grade orchestration and multi-tenant cluster isolation. The hiring profile (10 engineers, half senior/principal) and project focus on GPU cluster automation, SDN, and dynamic scaling suggest they're solving for real-time capacity matching, not just trading UI. Pain-point clustering around financing and risk de-risking indicates the market model itself is the core product—connecting buyers who can't commit long-term to sellers with spare capacity.

Tech Stack 27 technologies

Core StackLinux Rust Kubernetes Go Terraform C++ gRPC React Next.js Python CUDA NCCL InfiniBand KVM QEMU libvirt BGP VXLAN RoCEv2 GPU vLLM C SQL REST SLURM Weka PCIe

What San Francisco Compute Company Is Building

◆Challenges

Offloading customer risk
De-risking infrastructure build-out
De-risking gpu cluster financing
Creating liquid market for gpu offtake
Mitigating customer risk in gpu financing
Mitigating risk of gpu cluster development
Optimizing compute utilization
Risk of financing gpu clusters

▲Active Projects

Deploying new gpu clusters worldwide
Automating hardware deployments
Creating internal customer processes
Automation frameworks for provisioning and monitoring
Software-defined networking solutions
Building linux os image
Maintaining golden image
Designing api protocol
Orchestration frameworks for gpu clusters
Dynamic inference scaling systems

Hiring Activity

Steady10 roles · 4 in 30d

Department

Engineering

10

Product

1

Seniority

Mid

5

Senior

4

Principal

2

Company intelligence

Find more companies like San Francisco Compute Company by tech stack, pain points and active projects

Get started free

About San Francisco Compute Company

SF Compute is building a real-time trading venue for GPU compute, positioning compute as a liquid commodity traded between buyers and sellers. The company solves a supply-side problem: cluster operators can't fully book capacity, and buyers are forced into inflexible annual contracts. SF Compute's infrastructure spans global GPU cluster deployment, automated provisioning, and orchestration layers (Kubernetes, SLURM) paired with customer-facing trading and risk management tooling. The 11-person team is based in San Francisco and focused on core engineering and product work.

HeadquartersSan Francisco, California

Company Size11–50 employees

Hiring MarketsUnited States

Frequently Asked Questions

What tech does SF Compute use for GPU orchestration?

Kubernetes, SLURM, vLLM for inference, Rust/Go for systems code, and networking infrastructure (VXLAN, RoCEv2, InfiniBand, BGP) for multi-cluster communication and isolation.

What is SF Compute working on?

Global GPU cluster deployment, automated hardware provisioning, software-defined networking, dynamic inference scaling, and API/orchestration frameworks for cluster management.

Similar Companies in Technology, Information and Internet

Other companies in the same industry, closest in size

C

Candidata ai

Technology, Information and Internet

K

Kanz

Technology, Information and Internet

V

VIHU

Technology, Information and Internet