Vendor-agnostic Go SDK for integrating cloud providers into Brev's GPU compute platform.
Cloud providers implement a standard set of interfaces — instance lifecycle, instance types, networking, storage — and Brev handles orchestration, inventory sync, and end-user experience.
- Cloud providers integrating GPU compute into Brev
- NVIDIA Cloud Partners (NCPs) offering Brev-compatible infrastructure
- Compute brokers and marketplaces aggregating multi-cloud GPU capacity
Three resources to go from zero to a working provider integration:
- Cloud Manual — Complete reference for the SDK's interfaces, types, capabilities, billing, and provider expectations.
- Architecture — How the Cloud SDK connects to Brev's control plane: inventory sync, provisioning flow, and credential management.
- Integration Guide — Practical walkthrough to implement a new provider, with copy/paste scaffolding and validation tests.
- Security Requirements — Network model, SSH access, firewall rules, and encryption expectations.
- Validation Testing — Shared test suite for verifying provider implementations against real APIs.
- V1 Design Notes — Design decisions, known quirks, and AWS-inspired patterns in the v1 API.
- Version:
v1— internal interface, open-sourced - Cloud provider implementations are internal-only for now
v2will be shaped by feedback and real-world integration experience
We welcome and encourage contributions. If you're building GPU compute infrastructure or working on a provider integration, open an issue or submit a PR — we want to build this with the community.