generalist engineer · self-taught · runs his own lab

I build things on computers — and run most of them myself.

$ ▊

Firmware, servers, AI, payments, and the web — I've taught myself the whole stack by building it. Everything below is live and real, not mockups: the actual services running in my lab right now.

69
services running
3,685
MCP servers indexed
31
Bitcoin stores
8
sites self-publishing
01 · Capabilities

A broad toolbox.

I'm a generalist — I move between the OS layer, the network, the models, and the metal. Here's what I actually reach for.

Systems & infrastructure

ProxmoxDockerZFS · SLOG/L2ARC Linux · systemdnginxNetworking Cloudflare TunnelsBackups · PBS

AI & models

Ollama · multi-hostGPU passthrough Local LLMsFlash AttentionKV-cache quant MCP / agentsComfyUI

Software

PythonJavaScriptFlask React / ViteSQLREST APIs

Hardware & embedded

ESP32 · firmwareADB / AndroidMagisk Amlogic A/BBoot imagesRetroArch

Payments & security

Bitcoin · LightningBTCPaySelf-hosted payments Defensive securityOSINT
02 · Built & running

Things I've built — and still run.

Each one is linked and live. I didn't build these for a resume; I built them because I wanted them, and they're up 24/7.

Omninexus

mcp.thetempleofdoom.com
An MCP hub that indexes 3,685 AI servers — discovery, search, and health checks in one place.
why it matters Agents discover any capability instead of hard-coding integrations. One endpoint for thousands of tools.

Hyperion

hyperion.thetempleofdoom.com
An app store for AI agents — publish, discover, and call MCP tools, paid over Bitcoin.
why it matters Machine-to-machine billing: an agent pays for a tool the same way you'd buy an API key.

ARGUS

argus.thetempleofdoom.com
Rent-a-red-teamer — web-vuln scanning and OSINT for autonomous agents, running on Kali.
why it matters Turns a red-team engagement into a metered API call instead of a manual project.

Astraea

astraea.thetempleofdoom.com
A Washington divorce-law assistant with RAG over verified statute and DV-safety guardrails.
why it matters Answers grounded in real law — not a model hallucinating case law from memory.

The Temple

thetempleofdoom.com
A live index of all 69 self-hosted services, plus a provably-fair Bitcoin raffle.
why it matters One honest dashboard for an entire fleet — every card is a live service, not a screenshot.

Tarro

tarot.thetempleofdoom.com
A tarot reader that quotes Waite and Wilhelm/Baynes instead of guessing.
why it matters Knowledge apps that cite their sources — the model reads real texts, not vibes.

Nebula

nebula.thetempleofdoom.com
A paid image-generation GUI on a self-hosted ComfyUI stack.
why it matters Consumer-grade UI on raw diffusion models — no cloud, your own GPU.

Publisher fleet

ai · tech · science · crypto · linux · gaming · diy · guides
Eight authority sites that research, write, and self-publish — four posts a day each.
why it matters A content engine that runs itself, end to end, with no one at the wheel.

Multi-GPU LLM routing

3 GPUs + a MacBook · internal
Four GPU lanes and one model standard across three GPUs and a MacBook, routed by latency and VRAM.
why it matters Any app gets the right model on the right lane — no saturation, no cold-model timeouts.

And more: Evergreen Applicator (WA pesticide exam prep), Signal Miner (Telegram market intel), Rigel (no-KYC SOCKS5), LYNX (forensics API), Supernova (SMM) — the full list is at thetempleofdoom.com.

03 · Trajectory

How I got here.

Self-taught. I got into this by running my own servers and breaking them a lot.

2024 — now
Independent engineer
Built and run a homelab: a Proxmox fleet, GPU LLM routing, and a stack of self-hosted services and apps.
2023 — now
Field operations & logistics
Lime — operations in the field; honed the "diagnose and fix under pressure" mindset.
2014 — 2023
Master grower / controlled-environment agriculture
Managed precision environments — the same systems thinking I apply to hardware.
04 · Services & rates

What I can help with.

Self-hosted AI / GPU passthrough

$75/hr

Local LLM on your own hardware — GPU passthrough, Ollama setup, multi-host routing, context/VRAM optimization.

Homelab & sysadmin

$65/hr

Proxmox, Docker, ZFS, backups, networking. Turn a pile of servers into one clean infrastructure.

Hardware / embedded

$65/hr

Android TV rooting, ESP32 firmware, ADB automation, boot-image patching — the below-the-OS-layer stuff.

Defensive security review

$95/hr

Authorized vulnerability assessment and app-security review. Scoped, documented, lawful.

05 · Contact

Building something similar? Stuck on a local LLM or GPU setup?

Tell me what you're running and what's failing. Sometimes it's a two-command fix.