Personal project
Local AI. Real operations.
A local assistant that turns an infrastructure request into a working service across a three-node home cluster.
Repeated infrastructure tasks across three nodes.
Local inference → SSH and APIs → services.
Service provisioning and a CIFS fault diagnosis.
Published homelab example · Completion time not measured.
The problem
Managing containers, Docker services, reverse proxies and monitoring meant repeated manual work across several machines.
My contribution
I built a local AI chief-of-staff, connected it to my cluster through SSH, and used it to provision services and create a monitoring dashboard for my infrastructure.
Architecture choices
Local inference
Models run on my own hardware, alongside the infrastructure they manage.
Infrastructure access
The assistant uses SSH and application APIs to carry out administrative tasks.
Task-specific monitoring
The dashboard focuses on container status, storage, traffic, certificates and SSH activity.
What it showed
Service provisioning
The published example covers container creation, Docker deployment, a TLS proxy route and application setup.
Fault diagnosis
The assistant identified a CIFS metadata caching issue behind a storage problem.
Status & next evaluation
Personal homelab project. The published scope covers service provisioning, monitoring and infrastructure diagnosis.
A useful next evaluation is to record completion time, successful runs and interventions across a fixed set of infrastructure tasks.
Project source
The project details and reported results come from my linked LinkedIn publication.
Read the original project post