About
I run an AI-first engineering org. Eight years out of the editor. Then AI made building worth it again. Specs in, production out, nobody types code. Everything here follows from that shift.
Eight years without writing a line of code. I ran engineering orgs instead: platform, data, two acquisitions, five countries.
Then AI changed the math. 15,000+ GitHub contributions since March, and I still don't type code. I architect, spec, and review. Agents do the typing.
Now the playbook runs without me. Engineers improve it daily; a talent acquisition lead and a product designer ship to production on their own. The badge says Director of Engineering. The job is AI architect.
Capabilities
AI changed what one person can carry. This is the current surface area.
Spec before code, plan before agents, retrospective after every build. The method was forged in failures, including 50,000 CI minutes gone in five days, and it now ships production systems daily without me typing code.
Receipt 15,000+ GitHub contributions since March, every one AI-written, shipping to production daily.
The playbook is open to everyone at Evinced: engineers run it and make it better, and a talent acquisition lead and a product designer ship to production on their own. Who gets to build just changed.
Receipt A talent acquisition lead and a product designer ship to production on their own, no engineer in the loop.
My own ops platform spanning Jira, ClickUp, Zendesk, HubSpot, GCP, LLM routing, KPIs, and FinOps. Around it: a transcript platform, a public showcase, and the Tyr CLI family.
Receipt Empty repo to daily driver in under two months, spanning Jira, ClickUp, Zendesk, HubSpot, GCP, LLM routing, KPIs, and FinOps.
Token spend from $20K a month down to about $30 for 12 days. Idle GCP resources hunted and terminated. Tokens now do the work of two DevOps engineers, with budget left over for fun.
Receipt Token spend cut from $20K a month to about $30 for 12 days, while tokens do the work of two DevOps engineers.
GCP infrastructure and IT routines built so non-technical teams operate on their own: provision, deploy, and run without waiting on an engineer.
Receipt Non-technical teams provision, deploy, and operate GCP infrastructure end to end without waiting on an engineer.
Twenty years underneath all of this: event-driven systems at 50M+ sessions a day, multi-cloud Kubernetes, SOC2 and FedRAMP, and distributed orgs across five countries and two acquisitions.
Receipt Event-driven platforms at 50M+ sessions a day, through two acquisitions, with teams across five countries.
The arc · twenty years, one ladder
Selected work
Took 2 teams and 11 people to 7 teams and 40+ across five countries, stood up GenAI, ML, Data Infra, and DevOps from zero, and cleared FedRAMP readiness in six months to unblock a $10M government pipeline.
Deployed MCP-integrated LLM agents that automate triage, docs, and first-pass code review, plus multi-LLM workflows that dispatch Claude, Codex, and Gemini in parallel for reviews and plans.
Ran a 30-person org across four countries, re-architected anomaly detection to cut false positives 85% and time-to-insight from 12 hours to minutes, and shipped data integrations that influenced ~$12M in revenue.