Skip to content

Lab

This is the tooling and infrastructure I run day-to-day - the homelab, GPU workstations, live show-control rig, and edge hardware behind the projects. It is where ideas get prototyped, models get served on-prem, and events get run end-to-end.

Compute & virtualization

A two-node Proxmox cluster I treat as my private cloud - everything is containerised, versioned, and reproducible.

  • Proxmox Cluster

    Two physical nodes running VMs and LXC containers, with snapshots and live migration between hosts.

  • LXC + Docker (Dockge)

    Lightweight containers managed through Dockge for a clean, declarative stack view.

  • Self-Hosted Services

    Internal tooling, databases, and dashboards kept on-prem and under my control.

  • Mini-PC Nodes

    Compact NUC-class machines - an ASUS NUC 14 Pro and a Beelink 6600H - for low-power, always-on edge compute.

GPU & AI

On-prem GPU workstations that keep generation and inference fast, private, and off the meter.

  • RTX 4090 / 5090 Workstations

    High-VRAM cards for rendering, training runs, and real-time inference.

  • ComfyUI Render Stacks

    Node-based generation pipelines for image and video work.

  • On-Prem CUDA Inference

    Multi-model computer-vision pipelines served locally on CUDA - no data leaves the building.

  • Facial Recognition

    CompreFace and InsightFace for self-hosted face detection, matching, and check-in.

Event & show control

The rig I run for live events - registration through to the visuals on the wall.

  • Registration & Kiosks

    Event registration, self-service kiosks, facial-recognition check-in, and on-site badge printing.

  • Resolume

    Live visuals and media playback for stages and LED walls.

  • Bitfocus Companion

    Custom Stream Deck button surfaces that drive the whole show from one panel.

  • OBSBot PTZ

    AI-tracking pan/tilt/zoom cameras controlled hands-free during a live show.

  • ArtNet / DMX

    Custom ESP32 ArtNet nodes that map DMX universes onto LED strips and TFT grids, driving lighting and stage effects alongside the visuals rig.

Edge & IoT

Hardware I design, print, and program for jobs that have to run at the edge.

  • ESP32 / ESPHome

    Microcontroller firmware for sensors, displays, and actuators.

  • Home Assistant

    The automation brain tying devices, schedules, and dashboards together.

  • Bambu Lab H2D

    Dual-nozzle, large-format 3D printing for custom enclosures and rack parts.

  • Mini Racks

    DeskPi T0 Plus and 3D-printed 10-inch mini racks for compact, portable node clusters.

  • Custom Enclosures & Racks

    Purpose-built 3D-printed housings and racks for edge appliances.

Networking

Connectivity I design to be resilient - from a private mesh across sites to the on-site rig that keeps an event online and printing.

  • Tailscale Mesh

    A WireGuard mesh that ties every node, site, and device into one private network - no exposed ports.

  • Event Networking

    Managed WiFi with cellular failover and link bonding, so registration and check-in never drop mid-event.

  • Print Networking

    Tablets, print servers, and tuned queues across the floor - optimised for the fastest possible badge-printing times.

  • Portable Network Rack

    A flight-ready rack of switching, print servers, and access points I bring on-site and rack up in minutes.

Cloud & edge

When something needs to be global, it lives on the edge - provisioned as code.

  • Cloudflare

    Pages, Workers, D1, KV, Access, and Turnstile for fast, secure edge apps.

  • AWS + Terraform

    Cloud infrastructure defined declaratively and deployed reproducibly.

  • Azure Functions

    Serverless compute for event-driven workloads and integrations.

Self-hosted stack

The services that keep the lab running - owned, not rented.

  • RustDesk

    Self-hosted remote desktop for secure access to lab machines.

  • CloudPanel

    Server and site management for self-hosted web apps.

  • Keygen KMS

    Self-hosted license and key management for my own software.

  • Monitoring

    Metrics, logs, and uptime checks across the cluster.