I spent years across infrastructure engineering, product management, and enterprise cloud advisory. My value was judgment — knowing which technologies to invest in, which architectural decisions would age well, and which AI initiatives would actually deliver ROI.
The advice was good. The relationships were strong. But at some point in the last three years, I started feeling a gap between what I was recommending and what I was actually doing. I was telling enterprises how to build AI systems. I wasn't building them.
That gap bothered me enough that I did something about it.
What Changed
The shift happened gradually, then all at once. The gradual part was years of watching AI tooling mature — from "you need a team of ML engineers and a GPU cluster" to "you can build a useful RAG prototype with accessible tools." The all-at-once part was realizing, mid-way through an advisory engagement, that I could move from scoping the system to building a working version myself.
Not because I'm a better engineer. I'm not — there are engineers I've worked with who are exceptional at a depth I don't have. But because I understood the problem from the customer's perspective, the architecture from the platform perspective, and the deployment constraints from the enterprise perspective simultaneously. That combination — accumulated across engineering, product, and cloud roles — turned out to be exactly what building AI products requires.
What Building Taught Me That Advising Didn't
The most important thing I learned from actually building: the gap between "it works in demo" and "it works in production" is larger and different than I understood from the outside.
Advising, you hear about production issues secondhand. The customer tells you the deployment is struggling, you ask questions, you diagnose, you recommend. The feedback loop is long and indirect.
Building, the feedback loop is immediate and unforgiving. The agent you were proud of fails on an edge case you didn't think to test. The latency you didn't worry about in development becomes a UX dealbreaker when real users try it. The prompt that worked perfectly in evaluation behaves strangely when users phrase things you didn't anticipate.
That direct feedback loop made me a better advisor almost immediately. I stopped recommending architectural patterns I hadn't personally stress-tested. I started having more concrete conversations about failure modes because I'd hit them myself. The credibility I had in advisory conversations shifted from "I've seen many clients do this" to "I built this and here's what I found."
The Ventures That Came Out of It
I've shipped several things that started as personal projects and evolved into something more serious. Tier9 AI grew from repeated experiments with agent-orchestration scaffolding — I kept rebuilding the same foundations and eventually productized them. Other tools came from scratching my own itch: things I needed that didn't exist in a form that worked for how I operated.
None of these are the scale of the work I led at Amazon or AWS. They're not supposed to be. They're proof of work — evidence that I can close the loop from problem identification to a working implementation, not just the advisory portion of that arc.
What I'm Looking For Next
I want a role where the strategic and the technical are both required. Not "strategic leader who can talk to engineers" or "technical leader who can talk to executives." Actually both, in the same role, in the same week, on the same problem.
That role looks like: leading an AI product from roadmap to production, or building and running a team that does it, or serving as the connective tissue between an enterprise's AI ambitions and its execution capacity. The title matters less than the scope. The scope I'm looking for is end-to-end ownership of something that ships and that matters.
Experience across enterprise engineering, product management, and cloud advisory — combined with hands-on building — doesn't make me a startup founder. It makes me, I think, unusually useful to an organization that's serious about deploying AI at enterprise scale: someone who can think at the strategy level and get into the weeds when the system is misbehaving.
That's the role I'm wired for. That's what I'm looking for.
