Pakkit.net

Uses

The tools, stacks, and habits behind the work.

A living snapshot of what I build with, run on, and play with — from the development stack and homelab to the security defaults, creative rig, and competitive games that keep the curiosity loop running.

// USES editorterminallaptopservernotebookmiccontroller What I build, run, and play with

Build

Development stack

The day-to-day toolkit for turning ideas into shipped, typed systems.

Why Static-first and typed by default. Astro ships HTML instead of a framework tax, TypeScript catches the dumb mistakes before they reach a branch, and AI pair-builders take the boilerplate — while validation-first slices keep all three honest.

Why AI assistance doesn't get to erode the architecture →

Astro

Static-first, island-hydrated frontend. This very site runs on it.

  • Frontend
  • Islands
  • Content

TypeScript

Typed end to end — if it compiles clean, half the bugs never happen.

  • Types
  • DX
  • Safety

Laravel

Backend workhorse when an app needs real server logic and structure.

  • PHP
  • APIs
  • Backend

Docker

Everything builds and runs in containers so the host stays clean.

  • Containers
  • Reproducible
  • Isolation

GitHub

Source of truth, CI, and PR-driven review for every change.

  • Git
  • CI
  • PRs

Claude / Kiro

AI pair-builders for slices, refactors, and the boring boilerplate.

  • Agents
  • Pairing
  • Refactors

Validation-first slices

Thin vertical slices, each gated by check + build before it merges.

  • Slices
  • check
  • build

Run

Infrastructure & homelab

Where the services live — a private cloud steadily getting leaner.

Why The homelab is where I learned ops the hard way. A VMware/vSAN private cloud taught me what resilience actually costs; now I'm trading heavy VMs for lean containers so the stack is cheaper to run and easier to reason about at 2am.

Field notes from running my own private cloud →

VMware / vSAN

Years on a VMware + vSAN private cloud — the homelab that taught me ops.

  • Private cloud
  • Hyperconverged
  • Ops

Docker migration

Moving workloads off heavy VMs onto lean, right-sized containers.

  • Containers
  • Migration
  • Right-sizing

Cloudflare

DNS, edge, and tunnels sitting in front of self-hosted services.

  • DNS
  • Edge
  • Tunnels

PKI / cert distribution

Internal CA thinking: issue, rotate, and distribute certs without copy-paste.

  • PKI
  • Certs
  • Rotation

Monitoring & observability

Metrics, logs, and alerts so problems announce themselves first.

  • Metrics
  • Logs
  • Alerts

Defend

Security mindset

Defaults that make the safe path the easy path.

Why Security here isn't a layer bolted on at the end — it's the default posture. Secrets stay isolated, scopes stay small, and anything automated runs where I can see its blast radius. Trust boundaries should be obvious, not clever.

How this thinking shows up in client work →

Secrets isolation

Secrets live in vaults and env — never in the repo or a chat window.

  • Vaults
  • Env
  • No leaks

Least privilege

Every token, service, and human gets the smallest scope that works.

  • Scopes
  • RBAC
  • Minimal

Auditability

If it changed, there's a record of who, what, and when.

  • Logs
  • Trails
  • Accountability

Zero-trust patterns

Verify every request; trust no network by default.

  • Zero-trust
  • Identity
  • Verify

Safe automation

Automation runs in sandboxes with an explicit, reviewable blast radius.

  • Sandbox
  • Guardrails
  • Review

Create

Creative stack

The engineering brain pointed at sound and side experiments, as PakkitStorm.

Why Same engineering brain, pointed at sound and the occasional weird experiment — half performance, half R&D. It's where I prototype ideas with no pressure and let the good ones sneak back into the serious work.

What's currently on the bench in the lab →

DJing

Bass-forward, club-energy sets built to move a room.

  • EDM
  • Live mixes
  • Club

Set prep & mixing

Reading a room and building a night's arc, warm-up to afterhours.

  • Live mixes
  • Pacing
  • Club

AI experiments

After-hours tinkering with AI tools — sketching ideas fast, keeping the ones with taste.

  • AI
  • Generative
  • Experiments

Stream & content tooling

Overlays, scenes, and bots that make going live less of a chore.

  • Streaming
  • Overlays
  • Bots

Play

Gaming stack

Competitive habits that quietly sharpen the engineering work.

Why Overwatch is competitive practice for the instincts engineering needs: read the system, make the callout, review the tape afterward. The community tooling is mostly me automating game night so the fun scales without the chaos.

How gaming quietly taught me systems thinking →

Overwatch

Main competitive game — mechanics and game sense under pressure.

  • Competitive
  • Teamwork
  • FPS

Competitive mindset

Fast feedback, clear callouts, and a review after every match.

  • Feedback loops
  • Comms
  • Review

Community tooling

Discord bots and game-night infrastructure for the crew.

  • Discord bots
  • Automation
  • Events

Decision principles

Tools earn their place when…

I'm not loyal to any one tool — I'm loyal to a short list of properties. Something new has to clear this bar before it gets to stay.

  • they reduce friction — The boring parts get genuinely faster, not just rearranged into new busywork.

  • they make mistakes easier to catch — check + build fails loud and early, before a human has to notice the breakage.

  • they help future-me understand what happened — The commits, logs, and docs explain the decision long after I've forgotten it.

  • they keep secrets and trust boundaries clear — Where a credential lives is obvious, and crossing a boundary takes a deliberate step.

  • they let weird ideas become real without becoming chaos — A late-night prototype can grow up into a real system without a full rewrite.

Things I avoid

…and what doesn't make the cut

Nothing here is forbidden forever — these are just the patterns that have bitten me enough times to earn a default no.

  • Giant all-in-one rewrites — Big-bang changes hide where they broke. Thin slices fail small and obvious.

  • Tools that hide too much magic — If I can't reason about what it does, I can't debug it when it matters.

  • Workflows that only work on one machine — "Works on my machine" isn't a deployment strategy — it belongs in a container.

  • Undocumented production paths — If prod only exists in someone's head, it's already a future incident.

  • Automation with no audit trail — Automation I can't review or replay is just a faster way to be wrong.

This list pairs with what I'm into right now, and a lot of it earns its keep in the lab before it makes it into real work.