RSP Cloud
Documentation
Welcome to the ultimate AI infrastructure. Here you will find everything needed to provision NVIDIA H100s, manage petabytes of NVMe storage, and scale your AI APIs.
Getting Started
Start building your AI clusters using our Python SDK or direct API calls.
1. Install the SDK
2. Authentication
Set your API key in your environment variables to authenticate automatically.
Compute Instances
Deploy bare-metal NVIDIA B200, H100, and A100 GPU clusters seamlessly.
import rsp_cloud
# Yahan hum 8x H100 ka cluster bana rahe hain
client = rsp_cloud.Client()
cluster = client.compute.create(
gpu_type="H100_SXM5",
gpu_count=8,
image="ubuntu-22.04-pytorch"
)
print(cluster.status) # Output: 'provisioning'
High-Speed Storage
AI training requires massive data throughput. We provide Parallel NVMe volumes connected directly via GPUDirect Storage.
Block Storage (NVMe)
Up to 100 GB/s read speeds. Attach to any compute instance instantly for rapid checkpoint saving.
Object Storage (S3 API)
Store petabytes of datasets securely. Fully compatible with standard S3 protocols and AWS CLI.
VPC & Networking
Isolate your AI infrastructure securely. Every account comes with a default VPC.
-
✦
Quantum-2 InfiniBand
For multi-node clusters, instances communicate over 3.2 Tbps non-blocking fabric.
-
✦
Security Groups
Configure inbound/outbound firewall rules using JSON policies.
API Reference
Direct HTTP endpoints for managing your cloud ecosystem without the SDK.
Creates a new GPU instance.
curl -X POST https://api.rsplatforms.com/v1/compute/instances \
-H "Authorization: Bearer $KEY" \
-d '{"gpu_type": "B200", "count": 1}'
Lists all active NVMe storage volumes.
curl -X GET https://api.rsplatforms.com/v1/storage/volumes \
-H "Authorization: Bearer $KEY"
Enterprise Support
If you need custom VPC configurations or reserved B200 quotas, contact our engineering team.
Contact Sales Team