Cloud & Platform · 2026
SwarmOS Cloud Backend
Sandboxed cloud desktops for AI agents: spawn, command and kill on Google Cloud.
- My role
- Cloud lead · GCP sandboxing & security
- When
- 2026
- Status
- Hackathon · live on Cloud Run
- Stack
- Python, FastAPI, Cloud Run
01
Where it started
A multi-agent hackathon team needed isolated, observable environments where AI agents could run code and drive browsers without touching anyone's machine.
02
What I owned
Cloud lead · GCP sandboxing & security
03
How I built it
- Built and deployed a FastAPI control API on Cloud Run (London region) that spawns, commands and tears down agent workspaces.
- Packaged a workspace image (Python, Node, Playwright, Xvfb and noVNC for live visual streaming) in Artifact Registry, with GKE Autopilot pods for isolation.
- Hard-coded safety into every container, a 30-minute auto-kill and a destructive-command blocklist, all covered by a pytest suite.
04
What changed
The orchestration team got a live, documented five-endpoint API to spawn agent desktops on demand, with no infrastructure to manage.
PythonFastAPICloud RunGKE AutopilotDockerArtifact Registrypytest