$ neofetch --workshop
os: ubuntu lts · bare metal
cpu: server-class xeon · 14C/28T
ram: 64GB DDR4
gpu: gtx 1650 super · 4GB (active)
storage: ~3TB ssd + ~22TB hdd
duty: 24/7 · services over vms
workloads:
forecast-engine running
archive-dashboard running
agent-infra running
watchdogs armed
The nodebare metal
| CPU | Server-class Xeon · 14 cores / 28 threads |
| MEMORY | 64GB DDR4 |
| GPU | GTX 1650 SUPER · 4GB — auxiliary ML duty |
| STORAGE | ~3TB SSD tier + ~22TB HDD archive tier |
| OS | Ubuntu LTS — no hypervisor, no VM layer |
| DUTY CYCLE | 24/7 — long-running services, not experiments |
The GPU, honestly4gb is a feature
| ROLE | Embeddings · TTS · small models · vision helpers |
| NOT FOR | Frontier-scale reasoning — that rents by the token |
| WHY SMALL | The 4GB ceiling forces right-sizing instead of hoarding |
| STATUS | Working daily — see the VRAM budget discipline in Hermes |
Workload fleetalways on
| FORECAST ENGINE | Fish Bite — scheduled refresh, health-checked |
| AGENT INFRA | Hermes — cron autonomy, memory system, watchdogs |
| DASHBOARDS | Self-hosted tools serving daily use |
| POLICY | If a service exists, it self-hosts |
Why so leanconsolidated
| HISTORY | Hypervisor + big-VRAM era documented in the essay |
| LESSON | Capability integrity beats hardware throughput |
| NOW | Stable services on bare metal; heavy thinking delegated up-stack |
| FUTURE | Hardware follows the workload — not the other way around |
// why class-level specs
CPU class, RAM, and GPU class are the parts that matter for understanding the architecture — and they're on every resume anyway. Exact drive models, hostnames, and port maps are the parts that help exactly one kind of visitor. Those stay private; the reasoning is in the privacy note.