Platform engineer across macOS and Windows endpoints, running a self-hosted Proxmox homelab, and doing local LLM inference on consumer AMD GPU hardware. Everything here is either actively maintained or part of the current production environment.
macOS fleet management defined as code. OpenTofu (open-source Terraform) drives Jamf Pro configuration through a GitOps pipeline — policies, profiles, and scope changes go through version control before they hit production devices.
Active Directory administration and Windows endpoint engineering background. PowerShell scripting for fleet automation, GPO management, and cross-platform tooling that bridges the macOS and Windows environments.
Two-node Proxmox cluster running self-hosted services with zero external cloud dependencies for inference. NPM reverse proxy with Authelia SSO, internal DNS via Pihole, and direct PCIe GPU passthrough for local LLM workloads.
Local LLM inference on consumer AMD GPU hardware. Qwen 2.5 Coder 14B running at 8.4 GB VRAM via PCIe passthrough in a privileged LXC — no API keys, no rate limits, no data leaving the homelab.
Cross-platform scripting for fleet automation and local tooling. Zsh on macOS and Linux with custom aliases and functions for day-to-day platform engineering; PowerShell handles Windows fleet management and cross-platform automation tasks.