Architecture — System Design
Built on Microsoft Presidio, our architecture separates analysis, anonymization, and processing into dedicated services for maximum reliability.
Core Components
Analyzer Service
Detects PII using 317 custom regex recognizers (regex patterns) for structured data like IDs and tax numbers, plus spaCy/Stanza/XLM-RoBERTa NLP for names and locations. All models run on our own servers — no data is ever sent to third-party AI providers. Returns entity positions, types, and confidence scores.
Anonymizer Service
Applies your chosen transformation method. Supports Replace, Redact, Hash, Encrypt, Asymmetric Encrypt, Mask, and Keep.
Web Application
Next.js frontend with real-time processing and server-side rendering. Supports 48 languages with locale-aware routing. Responsive design for desktop, tablet, and mobile. Optimized for Core Web Vitals.
REST API
RESTful endpoints for programmatic access. JWT authentication with short-lived tokens. Rate limiting and quota management. Available via official SDKs for JavaScript and Python, or any HTTP client.
Data Flow
Documents flow through our services in a secure, auditable pipeline. Each service is independently deployed with health monitoring and automatic failover. All inter-service communication uses authenticated, encrypted channels.
- 1
Client sends document via HTTPS
- 2
Analyzer service detects PII entities
- 3
Results returned with positions and scores
- 4
Client selects anonymization method
- 5
Anonymizer applies transformation
- 6
Anonymized document returned to client