Skip to main content
Back to Work

AI-OS – Unified AI Operating System

Agentic Systems & AI Architecture
Built
AI Systems Architect

Architected and open-sourced a high-performance modular monolith system designed for agent routing, automated API key rotation, and real-time observability.

Dashboard
Platform screens1 / 12

Dashboard

Main system overview.

Overview

AI-OS is a high-performance modular monolith system designed for agent routing, automated API key rotation, and real-time observability.

The Problem

Managing multiple LLM APIs, handling rate limits, and monitoring agent performance can become extremely complex in production AI applications.

The Approach

Architected a unified AI operating system with a smart key manager, fallback models, and a FastAPI core engine for sub-10ms event routing.

Agent routing architecture

AI-OS manages high-performance agent routing, handling API limits and model fallbacks automatically to ensure reliable LLM integrations.

Incoming request → FastAPI core engine → Key rotation (Celery/Redis) → Claude Code Proxy → LLM response → Prometheus observability

My Contribution

Implemented a FastAPI core engine, a smart key manager using Celery, Redis, and Playwright, a Claude Code Proxy routing requests to fallback models (Gemini, Groq, OpenRouter), a Next.js admin dashboard, and Prometheus/Grafana observability pipelines. Containerized the entire 8-microservice stack using Docker Compose.

Key Capabilities

  • Sub-10ms event routing via FastAPI core engine
  • Automated API key rotation stack
  • Smart key management using Celery, Redis, and Playwright
  • Claude Code Proxy for fallback models (Gemini, Groq, OpenRouter)
  • Next.js admin dashboard
  • Prometheus/Grafana observability pipelines
  • Docker Compose containerization of 8-microservice stack

Technology Stack

FastAPI · Python · Celery · Redis · Docker

Project Context

Open-source high-performance agent routing and API management system.