Infrastructure
The GreenRide Vienna AI Platform
A modular, sovereign AI infrastructure for next-generation urban mobility.
Architecture Overview
Architecture Overview
The live stack is one Docker network behind nginx: Next.js for the website, FastAPI v7 (78 agents + boardroom), FastAPI v5 (SMLA mobility), FastAPI v57 (invoices and payments), PostgreSQL 16, and Redis 7. Inference is SMLA plus Ollama, Gemini, or DeepSeek — not a separate ONNX/TensorRT cluster.
Website + nginx gateway
↓
78 AI agents (FastAPI v7)
↓
SMLA mobility (v5) + invoicing (v57)
↓
PostgreSQL + Redis
On-Premise Deployment
Full stack deployable within city-owned data centers. No third-party cloud dependency.
Sovereign inference
SMLA mobility on v5, plus local Ollama or EU-reachable Gemini/DeepSeek. No raw city data is required for the LLM fallback.
Modular by Design
Enable only the modules your city needs. Pay for what you use.
Audit-Ready
Every inference, prediction, and decision is logged and explainable.
Technology Stack
Inference Layer
SMLA · Ollama · Gemini · DeepSeek
Orchestration
nginx · FastAPI (v7 / v5 / v57) · Next.js
Data Ingestion
REST APIs · agent run log · Redis cache
Storage
PostgreSQL 16 · Redis 7
Security
TLS · JWT · frontend key · bcrypt · RBAC
Observability
Loguru · Docker healthchecks · agent_runs audit