Indianapolis, Indiana

We run our own cloud.

self-hosted, highly-available infrastructure on hardware we own — and the custom systems, pipelines, and products that run on top of it. no VC, no serverless, no vendor lock-in.

see the architecture
● self-hosted · always on

the iron

every service runs on infrastructure we own and operate

edge · Cloudflare tunnels → Caddy / NPM · no open inbound ports
identity · Authentik HA SSO · forward-auth
compute · Proxmox cluster · LXC + KVM
data · Patroni HA Postgres · etcd + HAProxy
observability · Prometheus · Loki · Grafana · alerting

Compute notes →

a real cluster, not a single box

A 5-node Proxmox VE cluster running LXC + KVM, with live migration and snapshots. Current container count is live in the header.

Data read →

HA Postgres, at home

A 3-node Patroni cluster with etcd consensus and an HAProxy VIP. Automatic failover, kill-tested.

Identity notes →

one login for everything

Authentik HA SSO — redundant servers and outposts fronting services via forward-auth, on its own Postgres + Redis Sentinel.

Edge / Network read →

public without punching holes

Cloudflare tunnels mean zero open inbound ports; a Caddy/NPM hub reverse-proxies everything, with self-hosted DNS + ad-filtering.

Observability read →

we can see all of it

Prometheus + Alertmanager + Grafana with a Loki/Promtail log fleet and node_exporter across hosts, plus a custom health monitor that pages us.


built from scratch

systems we engineered ourselves — not products, not off-the-shelf

Memory read →

a persistent brain for our AI agents

Postgres source-of-truth → single-writer export to a git/Markdown mirror → hybrid retrieval (SQLite FTS5 keyword + pgvector semantic) served over a custom MCP server. Durable, searchable memory for stateless LLM sessions.

Agent hooks read →

24 hooks that make the agent smarter

A custom Claude Code hook framework across four lifecycle events — intent/complexity scoring, skill-gating, session-retrospective indexing, insight extraction, and an audit trail.

BigBrain notes →

one context API for every machine

A central service that syncs environment knowledge and the memory database across every workstation and container, so any agent boots with the same context.

Nap CLI + GitOps drift notes →

our own deploy toolchain

An in-house DevOps CLI (deploy, health, env-sync, changelog) plus a GitOps reconciler with live-vs-repo drift detection comparing a running /api/version against git HEAD.


what runs on it

production apps and tools — each one solved a real problem

Applications

Developer tools


uses

the toolset behind all of it

Languages

  • TypeScript
  • JavaScript
  • Python
  • Bash
  • Lua

Frontend

  • Next.js (App Router)
  • React
  • Tailwind
  • Vanilla HTML/CSS
  • Tauri · React Native

Backend / Data

  • Node · Fastify · Express
  • Drizzle ORM
  • PostgreSQL · SQLite · Turso
  • pgvector · FTS5
  • Redis

Infra / Ops

  • Proxmox · Docker · LXC
  • Patroni · etcd · HAProxy
  • Cloudflare · Caddy · NPM
  • PM2 · Prometheus · Loki
  • Authentik

AI / Agents

  • Claude (Anthropic)
  • MCP servers
  • Ollama (local)
  • RAG · pgvector
  • Agent hooks