About Me

I got into software because I wanted to make games. Still true. But somewhere along the way, the interesting problems moved. Not the graphics or the gameplay, but the systems underneath. How do you build multiplayer that doesn't fall apart? How do you set up a team so they can actually ship?

That's what I've been doing for 15 years. Build something hard. Make sure the team understands it. Help them own it. Get out of the way.

Started at SkyVu leading the Battle Bears team through 6 shipped games. I built SkyVuNetwork the backend that handled multiplayer, economies, clans, all the stuff that kept people coming back. Learned the hard way that networking is where games get really interesting (and where they break).

At Age of Learning, I built a framework that other teams could actually use. Not just fast, but clear. If you didn't design it, you should still understand it and own it. Watched engineers I'd worked with grow into leads. That part mattered more than shipping anything.

Jam City was about removing stupid bottlenecks. Game designers shouldn't need an engineer to tweak a level. So I built an editor tool that got out of the way. Let designers do what they're good at. The best platform work is work nobody notices because it just works.

At TallyUp, I built everything. Five games, tournaments, mobile and web. Started from scratch with four engineers. Bad decision early on? You feel it forever. Good decision? Saves you months. No safety net.

Now at Centerfield, I'm working on event processing at massive scale and figuring out how to actually use LLMs in production (turns out "use LLMs in production" is harder than "fine-tune a model"). Event pipelines handling millions of events a day. Observability so complex systems don't just become black boxes.

I want to build systems that actually work. Not just survive, but work well for the people using them. Games taught me that responsiveness matters. Distributed systems taught me that reliability at scale is hard. Both matter.

Everything else is just being useful. See the next problem before it becomes a disaster. Help the team own it. That's the job.