§ stack — parts list

The toolbox.

Six drawers of tools, what I use each one for, and where you can see it in my work.

Languages sketch

PART A-1 · FIG. 1

Languages

What I reach for first. Typed where it matters, fast where it counts.

TypeScript

since 2015

Default for web apps, APIs and tooling.

Go

since 2019

Services that need to be small, fast and boring.

Rust

since 2022

Systems work — the Karnelian kernel lives here.

Python

since 2017

ML, data pipelines and quick automation.

Solidity

since 2021

Smart contracts and Web3 experiments.

SQL

since 2011

Still the best language for asking data questions.

Frontend sketch

PART A-2 · FIG. 2

Frontend

Interfaces that load fast, read clearly and don't fight the user.

React

since 2016

The base for nearly every interface I build.

Next.js

Marketing sites, storefronts and dashboards.

TanStack

Router, Query and Start for full-stack apps — including this site.

Shopify Polaris

Embedded Shopify apps and merchant tools.

Tailwind

Design tokens and fast, consistent styling.

Framer Motion

Animation, like every drawing on this page.

Backend & data sketch

PART B-1 · FIG. 3

Backend & data

Where business rules meet storage. Multi-tenant by default.

Node.js

since 2014

APIs, workers and integrations.

PostgreSQL

My default database, with row-level security for tenants.

MongoDB Atlas

Document-heavy products and fast prototypes.

Redis

Queues, caching and rate limits.

Firebase

Realtime apps and mobile backends.

SAP HANA

Enterprise integrations for warehouse platforms.

Cloud & ops sketch

PART B-2 · FIG. 4

Cloud & ops

Infrastructure that is cheap to run and easy to hand off.

GCP / GKE

Kubernetes workloads and managed data services.

Azure

Enterprise clients and identity-heavy setups.

Docker

Everything ships in a container.

CI/CD

Automated tests and deploys — 150+ projects and counting.

Linux

Home turf since the teenage years.

VoIP

Phone systems for a national freight network.

AI & ML sketch

PART C-1 · FIG. 5

AI & ML

From audio classifiers in 2019 to agent systems today.

Agentic workflows

Multi-step agents with evaluation and review.

TensorFlow

Model training for classification work.

Keras

Fast model prototyping.

Librosa

Audio feature extraction for music ML.

Dialogflow

Early conversational agents, pre-LLM.

LLM tooling

Retrieval, tools, evals and guardrails.

Design & making sketch

PART C-2 · FIG. 6

Design & making

Pixels, polygons and solder. Building things with my hands keeps the software honest.

Figma

Wireframes, design systems and prototypes.

Unreal Engine 5

Building RamQuest.

Roblox Studio

Building Fable Shrine in Luau.

AutoCAD

Technical drawings and layouts.

Raspberry Pi & Arduino

IoT prototypes and dispatch hardware.

3D printing

Enclosures, mounts and the occasional toy.

Currently exploring

the drawer marked "experiments"
Rust async runtimesWebGPULocal LLMsUnreal Niagara VFXJapanese (the language, not a framework)

loading