ROOKWELD / THE WORKSHOP

Where systems
grow hands.

Physical computing, local infrastructure, strange interfaces and blueprints for machines that do more than live behind glass.

Enter the lab ↓
REALITYOS / HUDNODE 01

SYSTEM LINK

ONLINE
NODENULLFEATHER
NETTAILSCALE / OK
AILOCAL / READY
PING24 MS
ESP32-S3128×64 OLEDPROTOTYPE

01 / THE METHOD

Build small.
Connect carefully.
Learn from reality.

The workshop is where software meets resistance: unreliable networks, limited displays, physical inputs, constrained hardware and everything that makes a clean diagram become a real system.

Not every experiment becomes a product. Every one should make the next build more honest.

02 / BUILDS & PROTOTYPES

Things that have
left the screen.

01BUILT PROTOTYPE

RealityOS Pocket HUD

A compact ESP32-S3 endpoint that turns system state into a physical glanceable interface — built around a 128×64 OLED, Wi-Fi telemetry and alert overlays.

  • ESP32-S3 N16R8
  • SSD1306 OLED · 128×64
  • Wi-Fi / HTTP endpoint
  • INFO · WARN · CRIT overlays
02OPERATIONAL LAB

realitycore Homelab

An Ubuntu-based workshop server for services, monitoring, storage experiments and RealityOS infrastructure — designed to remain useful even while individual experiments change.

  • HP Z2 Mini G9
  • Ubuntu Server · Docker Compose
  • Prometheus · Grafana
  • Tailscale-connected
03BUILT / EVOLVING

Distributed AI Nodes

A split-compute approach where GPU-heavy local models live on the mobile workstation while web tools, services and monitoring remain on the server node.

  • ZBook Fury GPU node
  • Ollama local models
  • API-routed services
  • Private node network
04EXPERIMENT SERIES

Physical Interfaces

Small hardware explorations that test how software can sense, signal and respond beyond a screen — from presence and identity to light, sound and environmental input.

  • Raspberry Pi · Arduino
  • RFID · BLE · sensors
  • GPIO · SPI · I²C
  • LEDs · TTS · buttons

03 / SYSTEM BLUEPRINTS

Before the build,
draw the forces.

Selected architecture studies. Blueprints describe intent and system boundaries; they are not presented as proof that every connected component is currently production-ready.

BP-01HUD SIGNAL PATH
SYSTEM
STATE
→HUD
API
→ESP32
CLIENT
→OLED
OVERLAY

Turn distributed state into a compact physical status surface with bounded alerts.

BP-02DISTRIBUTED LAB
FURYGPU / MODELSPRIVATE MESHREALITYCORESERVICES / DATAEDGE NODESHUD / SENSORS

Place each workload where the hardware and availability make sense.

BP-03REALITYOS CONCEPT LAYERS
INTERFACELauncher · HUD · Quick views↓ORCHESTRATIONRouter · Specialist agents · Pipelines↓CONTINUITYMemory layers · Timeline · Incubator↓INFRASTRUCTUREModels · Tools · Nodes · Vault

A system-design reference for RealityOS. The full case study remains paused until the current implementation can be inspected and verified.

04 / ON THE BENCH

Next experiments.
Small by design.

A deliberately short queue. These are directions for the next useful build—not promises disguised as finished products.

B-01NEXT ITERATION

HUD Input Layer

Add a physical button path and intentional acknowledgement flow to the existing ESP32 status display.

B-02EXPERIMENT

Presence & Identity

Explore bounded BLE or RFID presence signals without turning the prototype into an unnecessary surveillance system.

B-03CONCEPT

Ambient Node

A small Raspberry Pi or ESP32 surface for light, sound and environmental signals connected to RealityOS services.

05 / MATERIALS & TOOLS

The stack changes.
The method survives.

COMPUTE

ESP32-S3 · Raspberry Pi · Arduino · HP Z2 · ZBook Fury

INTERFACES

OLED · RFID · BLE · GPIO · SPI · I²C · sensors · LEDs

SOFTWARE

Python · Flask · FastAPI · Node-RED · Docker · Ollama

OPERATIONS

Linux · Windows · Tailscale · Prometheus · Grafana

DESIGN

Blueprints · rapid prototypes · telemetry · failure-aware flows

THE WORKSHOP / ALWAYS IN PROGRESS

Useful things begin
as unreasonable experiments.

View selected work →Build something together ↗