🛡️ KVCHub — On-Device Privacy & High-Speed Neural Intelligence
Welcome to the official KVCHub Organization profile!
At KVCHub, we research and build ultra-lightweight, hardware-accelerated, zero-trust AI models. Our mission is to make advanced privacy redaction and neural intelligence run 100% locally on-device with zero cloud data leakage, zero subscription fees, and sub-millisecond execution speeds.
🚀 Flagship Models
| Model | Parameters | Weight Size | Latency | Target Hardware | Hugging Face Link |
|---|---|---|---|---|---|
| Kava Privacy 2.01M | 2.01M |
8.0 MB | 2.2 ms | Apple Silicon GPU (MLX) & Workstations | KVCHub/Kava-Privacy-2M-mlx |
| Kava Privacy 636k | 636k |
2.5 MB | < 1.0 ms | Mobile, Edge, Embedded, & Streaming | KVCHub/Kava-Privacy-636k-mlx |
🌟 Core Features & Breakthroughs
- 🔒 Zero Data Leakage (100% On-Device): Text never leaves local memory. Fully compliant with HIPAA, GDPR, and SOC2.
- ⚡ Sub-Millisecond Speed: Runs up to 130x faster than cloud network roundtrips to OpenAI or Google DLP.
- 🧪 Zero False Positives: Immunized against scientific constants (
299,792,458 m/s), subnet masks (255.255.255.0), legal rules (12(b)(6)), and stock tickers ($182.50). - 🌍 Zero-Shot International: Detects French, UK, German, and Japanese addresses, phones, and names zero-shot.
- 🌊 Real-Time Streaming: Sanitizes live streaming LLM outputs token-by-token on the fly.
⚡ Quickstart
Install the official Python package from PyPI:
pip install kava-privacy
Run PII redaction in 2 lines of Python:
from kava_privacy import KavaPrivacy
# Initialize KVCHub Privacy Engine
kp = KavaPrivacy(model_name="2.01M")
redacted_text, entities = kp.redact("Email me at jane.doe@gmail.com or call 555-123-4567.")
print(redacted_text)
# Output: "Email me at [EMAIL] or call [PHONE]."
Sanitize entire CSV, JSON, or Log files via CLI:
python3 kava_clean.py customer_data.csv -o clean_data.csv
📜 Principles & Open Source
All KVCHub models and software are released under the permissive MIT License. Free for commercial and open-source applications worldwide.
- PyPI Package: pypi.org/project/kava-privacy/
- Hugging Face Org: huggingface.co/KVCHub
Built with ❤️ by the KVCHub Engineering Team