Open to Head of AI · VP AI Engineering · Field CTO · AI Transformation · AI Modernization · Executive advisory
The tools landed. Individual engineers got faster. Nothing downstream moved, because review, QA, compliance, and release were sized for the old throughput and nobody resized them. Then the harder question arrives from the board: can you prove any of this output is correct?
Answering that is an executive job. I have spent twenty-five years on the engineering side of it, and the last two years writing down exactly how it goes wrong.
What gets escalated as an AI problem is usually a capacity problem one step downstream. The three steps to the right walk you through it in forty seconds.
Any candidate can describe the problem. This is what I would do about it, starting day one. Ask me to defend any of it.
Where AI is already in use, including the shadow usage nobody put on a slide. What it produces, who reviews it, and whether that review could catch an error if there were one.
Take the highest volume workflow and put real evaluation behind it. Deterministic checks, a test-backed gate, telemetry. Publish the numbers internally, including the unflattering ones.
Cost-aware model routing with a budget leadership can see, an evaluation harness the teams own rather than rent, and an escalation policy where “a human reviews it” names a person and a threshold.
Four rules, in the order I would apply them.
Fix the constraint that actually binds. Almost never the model, almost always review and release.
Move at the speed people can absorb. The rollouts that stick are the ones that took the humans seriously.
Govern for verification, not for confidence. If nobody can say what would falsify the output, it is not governed.
Track the capability the organization is quietly outsourcing. Dependence compounds like any other debt.
Most candidates for a role like this ask you to take their judgment on faith. Mine is on the record, written in public and dated. Plotted by when, and by how long it took to make the case.
The writing is recent. The judgment behind it is not.
Fifteen years at Growth Acceleration Partners as Principal AI Solutions Architect and team lead, advising executive, sales, product, and engineering stakeholders on modernization strategy, feasibility, and risk, while building the multi-agent systems that did the work.
Before that, Director of Web Software Engineering at ITS on the senior management team, presenting at industry conferences on web architecture and high performance computing. Earlier, engineering leadership at World Travel Holdings running private-label e-commerce for top global travel brands across teams in Massachusetts, Florida, and Brazil.
Formally trained in biochemistry and philosophy, which is a strange background for this work right up until the moment someone has to decide what counts as evidence. Both are disciplines about what you are entitled to claim and on what grounds. That turns out to be most of this job.