Multi-persona code review
Code review that adds engineering, security and product perspectives in a single pass. For teams that need senior-level review without senior-level bottlenecks.
Faster reviews. Fewer post-merge surprises.Builder and operator. I ship software, and I lead the programs it lives in.
Fifteen years running programs at companies rebuilding themselves. Currently leading a multi-system integration platform, while shipping AI tools and consumer software alongside.
Start a conversation →I lead complex programs and I ship the software that runs alongside them. Both, on purpose. The combination is rarer than it should be.
The program work has been at companies rebuilding themselves. A national financial services group launching mobile and digital products across hundreds of locations in Canada and the US. A fintech replacing call-centre-dependent operations with self-serve software, growing sales 61.5% before acquisition. A marketing technology firm moving clients off agency dependency.
The build work spans internal tools and consumer products. Custom AI agents that route across Claude, Gemini and GPT, paired with n8n, Python and Apps Script. Full-stack consumer applications using React, Next.js and Postgres. The point is never the technology. The point is whether the tool removes friction from the work.
I write about what I learn for operators and leaders trying to make AI earn its keep.
Code review that adds engineering, security and product perspectives in a single pass. For teams that need senior-level review without senior-level bottlenecks.
Faster reviews. Fewer post-merge surprises.Delivery margin visibility for consulting organizations. Pulls from time tracking, project plans and rate cards. Shows project profitability in near-real-time, no analyst needed.
Margin conversations move from monthly to weekly.An overnight briefing system. Covers what changed, what's at risk, and what needs a decision before the first meeting of the day.
Less inbox triage. More leverage on the work that matters.Custom agents that route between Claude, Gemini and GPT based on what each model does best. Built for the kinds of work operators actually do, including a case study agent and several others.
Right model for the task. Less manual switching.A solo full-stack consumer build. Authentication, data modeling, multi-step user flows. Built end to end to validate a product hypothesis and prove the patterns now reused on the next build.
Patterns validated. Foundation locked in.A second consumer product in active build. Same end-to-end ownership applied to a new problem. Component system from scratch, iterative shipping, real production infrastructure.
In flight.Weekly notes on what works, what doesn't, and how to tell the difference. Written from the middle of an active enterprise integration program, not from the cheap seats.
Read on Medium →