AI/ML
HumanSign
A privacy-preserving authorship verification system using behavioral biometrics and ML classifiers.
Tech Stack
PythonScikit-learnFastAPINext.js
The Problem
Distinguish human-typed content from AI-generated or copy-pasted text using behavioral biometrics without collecting personally identifiable information.
How It Works
Captures keystroke dynamics in the browser, sends encrypted logs to a FastAPI backend. A trained Random Forest classifier analyses the patterns and returns an authorship confidence score. SHA-256 log chaining ensures tamper-proof audit trails.