ood-braket-adapter

README

ood-braket-adapter

Open OnDemand adapter for Amazon Braket — submit, monitor, and cancel quantum tasks on QPUs and managed simulators from the OOD portal.

Part of the aws-openondemand ecosystem.

What it does

Translates OOD job submissions into Amazon Braket quantum tasks. Users write circuits in OpenQASM 3.0 or Python (via PennyLane/Qiskit), submit them through OOD, and get results back in their home directory as S3-backed files.

Supported backends:

Device	ARN prefix	Notes
SV1 (state vector simulator)	arn:aws:braket:::device/quantum-simulator/amazon/sv1	34 qubits, exact
TN1 (tensor network simulator)	arn:aws:braket:::device/quantum-simulator/amazon/tn1	50 qubits, large circuits
DM1 (density matrix simulator)	arn:aws:braket:::device/quantum-simulator/amazon/dm1	17 qubits, noisy
IonQ Aria	arn:aws:braket:us-east-1::device/qpu/ionq/Aria-1	Real QPU
Rigetti Ankaa	arn:aws:braket:us-west-1::device/qpu/rigetti/Ankaa-9Q-3	Real QPU

Installation

go install github.com/scttfrdmn/ood-braket-adapter@latest

Or download a pre-built binary from Releases.

Usage

submit

Reads a JSON job spec from stdin, submits a quantum task, and prints the task ARN.

echo '{
  "circuit": "OPENQASM 3.0; qubit[2] q; h q[0]; cnot q[0], q[1];",
  "device": "arn:aws:braket:::device/quantum-simulator/amazon/sv1",
  "shots": 1000,
  "output_bucket": "my-braket-results",
  "job_name": "bell-state"
}' | ood-braket-adapter submit --region us-east-1
# arn:aws:braket:us-east-1:123456789012:quantum-task/abc123...

status

ood-braket-adapter status arn:aws:braket:us-east-1:123456789012:quantum-task/abc123
# {"id":"arn:...","status":"completed"}

delete

Cancels a quantum task (only possible while QUEUED or RUNNING):

ood-braket-adapter delete arn:aws:braket:us-east-1:123456789012:quantum-task/abc123

Job spec fields

Field	Type	Required	Description
circuit	string	yes	OpenQASM 3.0 circuit or Braket IR JSON
device	string	yes*	Braket device ARN (* or use --device flag)
shots	int	no	Measurement shots (default: 1000)
output_bucket	string	yes*	S3 bucket for results (* or use --output-bucket)
job_name	string	no	Display name in OOD job table

IAM requirements

The OOD instance role needs:

{
  "Effect": "Allow",
  "Action": [
    "braket:CreateQuantumTask",
    "braket:GetQuantumTask",
    "braket:CancelQuantumTask",
    "braket:SearchQuantumTasks"
  ],
  "Resource": "*"
}

Plus S3 write access to the output bucket.

OOD cluster configuration

# /etc/ood/config/clusters.d/braket.yml
v2:
  metadata:
    title: Amazon Braket
  job:
    adapter: script
    submit_host: localhost
    bin: /usr/local/bin/ood-braket-adapter
    submit:
      script: |
        #!/bin/bash
        echo "$OOD_JOB_SPEC" | /usr/local/bin/ood-braket-adapter submit \
          --region "$AWS_REGION" \
          --device "$BRAKET_DEVICE_ARN" \
          --output-bucket "$BRAKET_OUTPUT_BUCKET"