NanoFID | Terms of Service

Last Updated: June 24, 2026

1. Nature of the Service

NanoFID is an advanced facial recognition and machine learning research platform. By using this application, you acknowledge that NanoFID is an experimental tool designed to evaluate algorithmic bias, model generalization, and the mathematical foundations of identity verification. It is not intended for high-stakes security, law enforcement, or surveillance applications.

2. User Authentication and Security

Authentication: NanoFID utilizes Google Sign-In via Supabase. We implement asymmetric ES256 JWT (JSON Web Token) authentication. This ensures that the backend verifies your identity using public-key cryptography without ever storing or accessing your primary Google credentials.

Authorized Use: You agree to use the camera interface and image upload functionality only for your own face, or for individuals who have provided explicit verbal or written consent to participate in your research datasets. You may not use NanoFID to identify, surveil, or collect biometric data on individuals without their knowledge and consent.

3. Data Processing and Embeddings-Only Architecture

NanoFID is built on a privacy-first architecture. To protect user identity, we adhere to the following technical constraints:

Volatile Processing: Raw images uploaded for training or captured via Live Mode WebSocket are processed in-memory using the InsightFace stack.

Immediate Discard: Once a 512-dimensional embedding (a numerical vector representation) is extracted, the source image is immediately purged from our systems. NanoFID does not store raw photos or video frames.

Vector Storage: We store only the L2-normalized vector embeddings within a secure Postgres database using pgvector. These mathematical coordinates cannot be easily reversed to recreate your original likeness.

4. Live Inference Mode

By engaging Live Mode, you initiate a real-time WebSocket connection.

Frame Handling: The system processes frames at the edge and performs inference on our Python-based backend.

Performance Disclosures: To ensure stability on CPU-based hosting, we utilize the buffalo_sc model. You acknowledge that this provides a balance of 30fps performance with a nominal trade-off in absolute verification accuracy compared to larger models. NanoFID is a research tool and makes no warranty as to the accuracy of any verification result.

Third-Party Consent: If you direct the camera at any individual other than yourself during Live Mode, you are solely responsible for obtaining that individual's consent. NanoFID disclaims all liability for use of the camera interface on non-consenting third parties.

5. Limitation of Liability

NanoFID is provided "as is" and "as available" without warranties of any kind, express or implied. To the fullest extent permitted by law, the developer shall not be liable for any indirect, incidental, or consequential damages arising from your use of the platform.

6. Termination

We reserve the right to suspend or terminate access to NanoFID at any time, with or without notice, for conduct that violates these Terms or that we determine to be harmful to other users, the platform, or third parties.

7. Governing Law

These Terms shall be governed by and construed in accordance with applicable law. Any disputes shall be resolved through good-faith negotiation before pursuing formal legal proceedings.

8. Contact

For inquiries regarding the ML pipeline or data privacy, please contact the lead developer Oleh Lahoda at: qci.research@gmail.com