The central knowledge base, equipment inventory, service documentation, and infrastructure repository for AIT Brainlab (Asian Institute of Technology).
┌─────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ AIT BRAINLAB ARCHITECTURE │
└─────────────────────────────────────────────────────────────────────────────────────────────────────────┘
│
┌───────────────────────────┬───────────────┴───────────────┬───────────────────────────┐
▼ ▼ ▼
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ CORE MGMT │ │ SERVICES │ │ RUNBOOKS & DOCS │
│ [`mgmt/`](mgmt/)│ │[`services/`](...)│ │ [`docs/`](docs/)│
├─────────────────┤ ├─────────────────┤ ├─────────────────┤
│ • ait-brainlab- │ │ • JupyterHub │ │ • Infra Runbooks│
│ mgmt ($5/mo) │ │ • Web Print │ │ • Onboarding │
│ • Cloud DNS │ │ • Dual A6000 GPU│ │ • Offboarding │
│ • NetBird Mesh │ │ • DockerSpawner │ │ • Troubleshooting
│ • lldap Directory│ │ │ │ • Admin Roles │
└─────────────────┘ └─────────────────┘ └─────────────────┘
The decoupled, permanent control plane running under GCP project ait-brainlab-mgmt (~$0.45 to $7.45/month):
- Master Roadmap & Next Steps: Core management operations tracker and upcoming service deliverables.
- Identity-as-Code: Declarative
members.yamland GraphQL user synchronizer. - Network-as-Code: Declarative NetBird Software-Defined Networks (
network.yaml). - Modular Terraform IaC: Consolidated 2-layer Terraform (
foundation/,vm/).
services/jupyterhub/: Multi-user GPU JupyterLab container environment onla(dual RTX A6000, TrueNAS NFS).services/printing/: Remote Web Print Portal (docker-cups) bridging cloud to CSIM printer.
- Infrastructure & Server Runbooks: Ubuntu 22.04, NVIDIA CUDA, TrueNAS NFS/iSCSI, SSSD, and 1-command bootstrap.
- Network & Mesh VPN: NetBird mesh VPN setup, CSIM proxy configuration, and DNS topology.
- Research Cloud Workloads: Spot GPU templates, GCS buckets, and research grants ($5k/$1k).
- User Onboarding: Step-by-step SOP for new researchers and students.
- Member Offboarding: Data archiving and access revocation SOP.
- SysAdmin Troubleshooting: Incident runbook for CUDA, NFS, Proxy, and container failures.
- Roles & Handover: Matrix of responsibilities for Infrastructure Admins and Service Admins.
Historical configurations, legacy Docker images, screenshots, and older notebooks preserved for reference.
This repository uses uv for blazing-fast, deterministic Python dependency management:
# 1. Sync virtual environment and install all dependencies
uv sync
# 2. Run Identity GitOps synchronization
uv run mgmt/identity/sync_users.py
# 3. Run NetBird VPN GitOps synchronization
uv run mgmt/vpn/sync_netbird.py
# 4. Run Day 1 Ansible mesh operations playbooks
uv run ansible-playbook -i mgmt/ansible/inventory.ini mgmt/ansible/enroll_netbird.yml- No Hardcoded Secrets: Never commit passwords, private keys, or API tokens to version control.
- Proxy Aware: Outbound traffic on CSIM network requires
http://192.41.170.82:3128. - Persistent User Data: Always stored on TrueNAS at
/mnt/pool-1/home/{username}.