Platform

Storage

BRICQS Storage provides managed object storage buckets for AI artifacts — training data, model weights, embeddings, and output files — accessible from your deployments with zero-config authentication.

Creating a bucket

bricqs storage create my-dataset-bucket --project my-project

Buckets are provisioned in the same region as your deployments (westus2 currently) for low-latency access from your containers.

Uploading files

bash# Upload a file
bricqs storage upload my-dataset-bucket ./data/train.jsonl

# Upload a directory
bricqs storage upload my-dataset-bucket ./data/ --recursive

Or upload directly using the connection string shown in the console:

bricqs storage upload-batch my-dataset-bucket ./data/ --recursive

Accessing storage from a deployment

Inject the bucket connection string as a secret and read it in your container:

pythonimport os
from azure.storage.blob import BlobServiceClient

client = BlobServiceClient.from_connection_string(
    os.environ["STORAGE_CONNECTION_STRING"]
)
container = client.get_container_client("my-dataset-bucket")

# Stream a blob
blob = container.get_blob_client("train.jsonl")
data = blob.download_blob().readall()

Listing and managing files

bash# List blobs in a bucket
bricqs storage ls my-dataset-bucket

# Download a file
bricqs storage download my-dataset-bucket train.jsonl ./local/train.jsonl

# Delete a file
bricqs storage rm my-dataset-bucket train.jsonl

Storage tiers

HotDefault tier. Optimized for frequently accessed data (training runs, inference artifacts).CoolLower cost for data accessed less than once per month (archived outputs, backups).ArchiveLowest cost, highest retrieval latency. For long-term data retention only.

Pricing

Storage is billed at standard BRICQS Storage rates for the westus2 region. Egress within the same region (to your deployments) is free. Cross-region egress is billed at standard rates.

Note: Storage costs are shown in your project's Usage dashboard alongside GPU compute hours.