Things I've worked on

I'm a big fan of Go. The slightly awkward 2009 Go programming language promo with Russ Cox hooked me. A few months later, I used Go in a college paper on concurrency. Before that, I wrote C for toy robots and school projects, and I picked up Python at work. Go felt like fresh air: simple abstractions, fast feedback, and enough control.

One of my first larger Go projects was an open-source Redis client. Redis uses a text protocol over a socket: send a command, get data back, or mutate state. Building the client taught me to measure first, profile before optimizing, and keep network code simple under load.

I've written Go professionally since 2012. At KoGaMa, I built performance-critical services with Michael, Christian, and Jacob. We scaled to millions of users with weekend and holiday peaks. Our model produced high traffic and low conversion, so cost mattered. I focused on simple systems, commodity hardware, and our own content delivery setup.

At GameClub, I worked with Oliver and Niels on backend services and automation. While Eli maintained a catalog of more than one hundred iOS and Android games, we built workflows that let a small team keep that catalog running on modern devices and keep shipping updates.

Today I work at a logistics company, where I lead development on a Postgres-backed platform. The codebase and operational demands have both grown over the years. I enjoy that tension: keeping systems fast and reliable while the business moves toward more sustainable freight.

Technical focus areas

I spend much of my study time on database internals: B+ trees, storage engines, indexes, and query behavior. My notes on database systems are my working notebook.

Performance work is a constant thread in my career, from debugging regressions to redesigning systems that hit growth limits. I like to start with rough system math, then validate with profiling and tracing.

I recently ran a workshop on LLMs and software development. My view is practical: assistant workflows are the most useful for me today, fully autonomous agents still break on edge cases, and most visible gains come from longer thinking plus tool loops. In daily work, I use AI to draft and explore, then move from plausibility to verification with execution, tests, and review.

Tinkering and projects

I tinker to learn. Some projects stay small, and some become tools other developers use.

• OpenTUI (GitHub): I contribute to OpenTUI, a terminal UI library used for OpenCode. ongoing
• OpenTUI Bench (GitHub): A benchmark site for OpenTUI that stores benchmark history, compares commits, and flags regressions. 2026
• Tiger Style: A website that outlines the core principles of Tiger Style, a coding philosophy centered on safety, performance, and developer experience. 2024
• Color system: A web app for building color systems, inspired by Lyft's ColorBox.io. Built in Vue.js. 2019
• Chat: A chat prototype with member channels and direct messages. Built with NATS Streaming, Go, and JavaScript. 2018
• Redis client for Go: Open-source Redis client written in Go. 2011
• Self-balancing robot: A two-wheeled autonomous balancing robot. 2009

How I work

I like teams with direct communication and short feedback loops. We get the best results when we write things down, challenge ideas early, and ship in small, reversible steps. I optimize for safety, performance, and developer experience.

• I pull people in early, while the problem is still messy.
• I prefer minimal, reversible changes over large rewrites.
• I use napkin math to estimate throughput, latency, and cost before building.
• I set explicit limits, keep control flow predictable, and make production behavior observable.