Platform Architecture
DonzHit.me operates as a distributed system designed for secure public contribution, edge safety caching, and localized, private AI inference. Click on any pipeline stage below to view technical specifications.
Secure Client Ingestion
Frontend ApplicationsUsers record traffic safety incidents via cross-platform apps built using Flutter. Before any media is transmitted, the client requests a secure session from the backend, bound to cryptographically attested device states.
- Android: Play Integrity API verifies a genuine, untampered device and boot state.
- iOS: App Attest binds keys directly to the physical Secure Enclave.
- Web: Compiled to high-performance WebAssembly (Wasm) with JS fallback. Utilizes reCAPTCHA v3 to validate interaction scoring and blocks bots, and uses Javascript Interop for browser-native video/audio players.
- XR Emulator: Standardized mock streaming structures for smart-glasses integrations.
- XR Client (Android): Attestation bound to Snapdragon Spaces XR headset telemetry.
System Topology & Spatial Vision
High-level topology mapping the ingestion flow from spatial clients (Android XR) to bare-metal NVIDIA RTX 5090 VLM inference clusters.
Core Infrastructure Pillars
No backend API routes are accessible without cryptographic validation. Cloud Armor filters requests at the network edge while App Check token verification ensures all clients are validated before read/write permissions are granted.
95% of public feed requests are absorbed by Cloud CDN at the edge. The Go proxy forces ETag revalidation for client JS files while allowing browsers to cache immutable assets, maintaining sub-millisecond feeds during viral traffic.
Firestore databases are logically partitioned. Enterprise licensees operate on isolated database instances and cloud storage buckets, ensuring zero cross-tenant contamination while maintaining a unified backend deployment stack.