Hi, I'm Dustin Liukkonen.
I'm a senior solutions architect at AWS, based in Boston. Before that, I spent years as a software engineer and architect, at startups and large enterprises alike. Now I work with companies across a lot of industries, helping them take ambitious systems — lately, AI agents — from prototype to something they can run in production.
This site is where I think out loud. I’ve written about most of this in shorter form on LinkedIn (opens in a new tab) for a while — this is the first place I’m giving it room to stretch out. A lot of it is about agentic AI: what it takes to make agents reliable, observable, and safe enough to trust with real work. I helped author the Well-Architected Agentic AI and Generative AI lenses, which cover a lot of the same ground — cost, security, observability, where one agent’s job ends and the next begins.
What ties it together is a preference for systems you can actually understand — where you can see who talks to whom, tell whether the thing works, and undo a change that didn’t. That gets harder with probabilistic systems, and harder still as organizations scale — both are what I keep coming back to. Seeing the same pains recur across very different companies is where these field notes come from.
My interest areas
- Agentic & generative AI — getting agents from prototype to production, and evaluating systems that don’t give the same answer twice.
- Architecture — designing systems that stay understandable as they grow.
- Scaling organizations — the drag that shows up as companies grow: innovation that stalls, delivery that slows, and decisions that take too long.
- Cloud — AWS, Well-Architected, and the operational reality behind it.
- Observability & delivery — knowing whether the thing works, and shipping changes you can undo.
The best way to follow along is the RSS feed or the blog. If something here is useful — or wrong — I’d like to hear it.