Thirty years of enterprise infrastructure craft, applied to the era of agents. The same patterns that worked across nearly every Fortune 100, sized for whoever's actually in the room.
The story
The pattern across three decades has been consistent: show up where the next era is taking shape, then make the transition real for the people who aren't already at the front of the line.
It started with an ISP in the early 2000s, putting broadband into school districts the carriers wouldn't touch. It became virtualization at Pierce College before "cloud" was a word. It became vRealize Automation at DynamicOps and VMware. It became IaC at Fortune 100 scale at SovLabs and CloudBolt. Most recently, it became the first SE seat at Firefly.ai, and now the AI infrastructure work at Privian.
Every step has had the same structure. A new technology shows up. The people at the front of the line move quickly. Everyone else needs somebody to translate it into their reality, same craft, right-sized for them. That's what Privian does.
Privian has built cloud infrastructure for nearly every Fortune 100. Privian has also stood up the back office for a five-person nonprofit. The scale is different. The problem (figuring out what the technology actually is and making it work for your specific situation) is the same.
The writing at Cookies Cookie Settings Echoes of the Machine is the public version of the engineering. Practitioner-level posts on AI infrastructure, agentic ops, retrieval, evals, the autonomy ladder, and the cloud shape underneath all of it. If you want to know how Privian thinks, that's where to look.
Based in Orlando, Florida. Most engagements are remote. On-site available across Central Florida.
Early 2000s
Co-founded an ISP putting broadband into school districts the big carriers wouldn't touch. Ran Pierce College's full data center transformation: virtualized 80% of workloads before "cloud" was a word.
2010
Early SE at DynamicOps. Built what became vRealize Automation. After the VMware acquisition: served as National Cloud Specialist, contributed to early CodeStream CI/CD design.
Mid 2010s
Founding SE at SovLabs. Office of the CTO at CloudBolt. IaC, Kubernetes, and cloud-native at scale across nearly every Fortune 100.
Recent
First SE at Firefly.ai. Founded Privian to build the AI infrastructure the hyperscalers don't sell to.
Now
Consulting on AI infrastructure, agentic systems, and cloud architecture. Writing practitioner-level posts at Echoes of the Machine.
How we work
Not a values page. Just the operating assumptions that show up in every engagement.
01
If at the end of the engagement you don't have code that runs, infrastructure that's provisioned, or a written playbook you can act on, we didn't do the job. Strategy is part of the work, not the whole of it.
02
Every engagement ends with documentation, runbooks, and a handoff session. You own what we built. If you want to run it without us after that, you can. We'll wish you well.
03
If your existing platform is fine, we'll tell you. If the AI feature you want to build doesn't make sense, we'll tell you. If a competitor is a better fit, we'll point you at them. We'd rather lose the work than build the wrong thing.
04
The patterns that work at Fortune 100 scale also work for a five-person team. They just need to be right-sized. The local bakery deserves the same thinking as the bank. We make that real.
05
Not every workload belongs in the cloud. We'll happily put a Mac Studio in your office for training, batch, and eval if that's what the math says, and explain exactly why.
06
Architecture decisions go in ADRs. Operational steps go in runbooks. Patterns we use across engagements go on the blog. Tribal knowledge is a tax on the next person, including future-you.
What we don't do
Saying no on purpose is part of how we stay good at the things we say yes to.
If the deliverable is a slide deck nobody can build from, we're the wrong people. We do strategy because we have to ship from it.
We engage as senior practitioners, not as bench resources. Our value is in judgment, not hours.
We don't take vendor referral fees. If we recommend a tool, it's because it's the right tool, not because someone's paying us to recommend it.
Every engagement has a defined output and a defined end. If we can't say what you'll have at the close, we shouldn't be starting.
Tell us what you're working on. Thirty minutes is usually enough to know.
Start a conversation →