Human In The Lead
Why the AI trust gap is a harness gap — and the ladder from "in the loop" to "in the lead". Presented to the DevOps Toronto community.
Distinguished Engineer at Capital One Canada. Twenty years translating executive strategy into distributed systems — customer identity, AI platforms, and the libraries underneath them, with audit controls attached.
I work at the seam between executive strategy and distributed systems. At Capital One Canada I drive technology quality, AI adoption at high quality and velocity, and security across engineering in a regulated financial services business.
Before that, at Questrade, I owned enterprise AI architecture and spent years on Customer IAM, building identity infrastructure that protects millions of accounts and multi-billion-dollar assets. I treat platforms as products: success metrics, adoption, unit economics, developer experience, not just uptime.
I also talk about this work publicly, mentor engineers across 50+ teams, and advise on everything from threat modeling to AI-augmented SDLC. My deepest technical roots are in .NET, and I consume AI tooling the way other people drink coffee.
Photography, 3D puzzles and scale models, gaming, anime. I like systems with parts you can touch.
Click a node to expand. Hover an edge to see how a skill carried forward.
A non-exhaustive map of systems I have built and/or owned. Click any card for context.
Architecture, DevOps and engineering leadership — at conferences, inside engineering orgs, and in workshops for founders and startup teams. I speak externally representing the companies I work for.
Everything linked, nothing behind a login.
Why the AI trust gap is a harness gap — and the ladder from "in the loop" to "in the lead". Presented to the DevOps Toronto community.
Developers use AI more and trust it less. The trust gap is a harness gap — the distance between what your seniors know and what your systems encode.
Both sides of the AI-coding debate are arguing the wrong problem. The shift isn't machines writing code — it's machines joining the SDLC.
Every metric gets gamed once people know it is watched. Why velocity has to be anchored in outcomes, not activity.
An AI version of me, grounded in what’s on this page. Ask about architecture decisions, how I work, or what I’m curious about.
Advisory, speaking, recruiting, or a good architecture problem.