I work as a Data Center Network Engineer with a focus on GPU host deploy: fabric and host networking, bare-metal image and hypervisor bring-up, and data-center operations in FOSS-heavy environments.
Day-to-day themes (employer-neutral):
- Collapsed-core and colo-style data-center networking, with an Arista focus at work
- NetBox deploy and upkeep as IPAM/DCIM source of truth; Snipe-IT for inventory
- Proxmox and Ceph for hypervisor and storage deployments; edge-node and facility build-out patterns
- Automation bias for repeatable ops (load-shed and site HA themes from prior network-admin work) without claiming a specific orchestration product by name unless it is on the CV
Outside production constraints I run a deliberate homelab documented as lab proof:
- Multi-node Proxmox VE (corosync + external QDevice), Ceph RBD, TrueNAS SCALE, and Proxmox Backup Server
- VLAN / multi-plane segmentation; Arista switching plus lab Juniper EX; virtualized Arista VxLAN fabric for switching tests (lab only)
- Nginx TLS reverse-proxy at the edge; NetBox IPAM/DCIM
- FreeIPA (Kerberos/LDAP) with Keycloak application SSO over OIDC
- Grafana, Prometheus, and Zabbix; Wazuh SIEM and Greenbone vulnerability scanning
- Self-hosted Git, OpenProject, and an agent console
- GPU host deploy and dedicated GPU guests for local LLM inference; NVIDIA AI certification path in progress (not claimed complete)
This site stays employer-neutral. No customer names, no private networks, no credentials, no resume PDFs.