Privacy & Data Protection Policy

1. Data Minimization & Zero-Leakage Architecture

QuantumX operates under strict data minimization principles. We do not collect, store, or sell personal identifiers or raw patient biological specimens. All computational workloads entering the preprocessing autoencoder are normalized into bounded feature vectors prior to quantum angle encoding.

2. Information We Process

• Authentication Data: Email address, hashed credentials (Argon2id/bcrypt), or Google OAuth tokens necessary for managing secure user sessions.
• Diagnostic Feature Tensors: De-identified numerical vectors representing continuous biometric markers (e.g., nuclear perimeter, area, concavity, or transcriptomic counts).
• Execution Provenance: Algorithmic runtime telemetry, including optimization loss curves, parameter-shift gradients, and QPU shot counts.

3. Quantum Cloud Transmission & IBM Runtime

When real hardware execution mode is selected, parameterized quantum circuits are transpiled to native basis gates (CX, Rz, SX) and transmitted to IBM Quantum Runtime endpoints over TLS 1.3 encrypted connections. Only abstract circuit instructions and rotation angles are communicated to physical cryostats; no clinical patient context is ever exposed to external QPU schedulers.

4. Cryptographic Receipt Provenance

Every inference request generates a verifiable SHA-256 cryptographic receipt. This digest encapsulates the model configuration, the exact random seed utilized in stratified k-fold splits, and the resulting gate ablation importance scores. These receipts allow research audits without requiring persistent retention of raw underlying feature matrices.

5. User Rights & Data Deletion

You hold the absolute right to purge your uploaded datasets, custom VQC checkpoints, and analytical logs from our encrypted databases at any time through the Settings dashboard or by dispatching a deletion request to our compliance team.