devqubit 0.1.7

Quantum experiment tracking library

devqubit

Local-first experiment tracking for quantum computing. Capture circuits, backend context, and configuration so runs are reproducible, comparable, and easy to share.

Status: Alpha – APIs may evolve in 0.x releases.

Why devqubit?

General-purpose experiment trackers (MLflow, Weights & Biases, DVC) are great for logging parameters, metrics, and artifacts. But quantum workloads often need extra structure that isn't first-class there by default: capturing what actually executed (program + compilation), where it executed (backend/device), and how it executed (runtime options).

Challenge	MLflow / W&B / DVC	devqubit
Circuit artifacts	Generic file logging	Automatic OpenQASM 3, and SDK-native formats (first-class)
Device context	Must be done manually	Automatic backend snapshots, calibration/noise context (first-class)
Reproducibility	Depends on what you choose to log	Automatic program + device + execution fingerprints to detect what changed
Result comparison	Metric/table-oriented comparisons	Distribution/structural/drift-aware diffs
Noise-aware verification	Require custom logic	Configurable policies with noise tolerance
Portable sharing	Artifact/version workflows exist	Self-contained run bundles (manifest + SHA-256 digests)

devqubit is quantum-first: the same circuit on different backends (or the same backend on different days) can produce different distributions - devqubit helps you track why.

Features

• Automatic circuit capture – QPY, OpenQASM 3, and native SDK formats
• SDK adapters – Qiskit, Qiskit Runtime, Amazon Braket, Cirq, PennyLane
• Content-addressable storage – deduplicated artifacts with SHA-256 digests
• Reproducibility fingerprints – detect changes in program, device, or configuration
• Run comparison – TVD analysis, structural diff, drift detection
• CI/CD verification – verify runs against baselines with configurable policies
• Portable bundles – export/import runs as self-contained ZIPs

Documentation

📚 https://devqubit.readthedocs.io

Installation

Requirements: Python 3.11+ (tested on 3.11–3.13)

pip install devqubit

# With SDK adapters
pip install "devqubit[qiskit]"          # Qiskit + Aer
pip install "devqubit[qiskit-runtime]"  # IBM Quantum Runtime
pip install "devqubit[braket]"          # Amazon Braket
pip install "devqubit[cirq]"            # Google Cirq
pip install "devqubit[pennylane]"       # PennyLane
pip install "devqubit[all]"             # All adapters

# With local web UI
pip install "devqubit[ui]"

Quick start

Track an experiment

from qiskit import QuantumCircuit
from qiskit_aer import AerSimulator
from devqubit import track

qc = QuantumCircuit(2)
qc.h(0)
qc.cx(0, 1)
qc.measure_all()

with track(project="bell-state", name="baseline-v1") as run:
    backend = run.wrap(AerSimulator())
    job = backend.run(qc, shots=1000)
    counts = job.result().get_counts()

    run.log_param("shots", 1000)
    run.log_metric("p00", counts.get("00", 0) / 1000)

print(f"Run saved: {run.run_id}")  # or use run.name

The adapter captures: circuit (QPY + QASM3), backend config, job metadata, and results.

Comparing runs

from devqubit.compare import diff

# By run name (with project context)
result = diff("baseline-v1", "experiment-v2", project="bell-state")

# Or by run ID
result = diff("01JD7X...", "01JD8Y...")

print(result.identical)           # False
print(result.program.match_mode)  # "structural"
print(result.tvd)                 # 0.023

Or via CLI:

devqubit diff baseline-v1 experiment-v2 --project bell-state

CI/CD verification

Verify that a run matches an established baseline:

from devqubit.compare import verify_baseline, VerifyPolicy

policy = VerifyPolicy(
    tvd_threshold=0.05,
    require_same_device=False,
)

result = verify_baseline(
    "nightly-run",  # run name or ID
    project="vqe-hydrogen",
    policy=policy,
)

assert result.ok, result.reason

In CI pipelines, use the CLI with JUnit output:

devqubit verify nightly-run --project vqe-hydrogen --junit results.xml

CLI reference

devqubit list                                       # List recent runs
devqubit show <run>                                 # Show run details
devqubit diff <run_a> <run_b> --project myproj      # Compare two runs
devqubit verify <run> --project myproj              # Verify against baseline
devqubit pack <run> -o bundle.zip --project myproj  # Export portable bundle
devqubit unpack bundle.zip                          # Import bundle
devqubit ui                                         # Start web UI

Note: <run> can be a run ID or run name. When using names, provide --project for disambiguation.

See CLI reference for full interface information.

Web UI

devqubit ui
# → http://127.0.0.1:8080

Browse runs, view artifacts, compare experiments, and set baselines.

Configuration

Environment variable	Default	Description
DEVQUBIT_HOME	~/.devqubit	Workspace directory
DEVQUBIT_CAPTURE_GIT	true	Capture git commit/branch/remote
DEVQUBIT_CAPTURE_PIP	true	Capture installed packages
DEVQUBIT_VALIDATE	true	Validate records against schema

See configuration guide for advanced options.

Project structure

devqubit/                    # Metapackage (re-exports from engine)
packages/
├── devqubit-engine/         # Core: tracking, storage, comparison, CLI
├── devqubit-ui/             # FastAPI web interface
├── devqubit-qiskit/         # Qiskit adapter
├── devqubit-qiskit-runtime/ # IBM Runtime adapter
├── devqubit-braket/         # Amazon Braket adapter
├── devqubit-cirq/           # Google Cirq adapter
└── devqubit-pennylane/      # PennyLane adapter

Contributing

1. Install uv
2. Clone and sync:

   git clone https://github.com/devqubit-labs/devqubit.git
   cd devqubit
   uv sync --all-packages --all-extras
   

3. Install hooks and run checks:

   uv run pre-commit install
   uv run pre-commit run --all-files
   uv run pytest
   

See CONTRIBUTING.md for full guidelines.

License

Apache 2.0 – see LICENSE.