Walk-in closet, from carcasses up.
Custom cabinets designed and built from scratch — cut lists, soft-close drawers, and a secret compartment behind a pivot-hinged door.
20 years in software, 10+ leading engineering at enterprise scale. The last two: leading an AI-native transformation — 10 engineers shipping 150+ PRs a week across a 50-repo platform.
Every org has the tools now. Most are still using AI as a faster chatbot — and wondering why nothing compounds.
The resistance isn't about output. It's identity-driven: people protecting the expertise that made them valuable. No tool purchase fixes that.
What fixes it is the operating model — every initiative tied to an outcome, tool choice left to engineers, and leaders who ship with the same tools they preach.
I've run this transformation once, end to end. I'm looking for a bigger org to run it again.
Not a tooling rollout — a redesign of how work gets done. Each step compounds the previous one, and the whole thing is measured in shipped outcomes, not sentiment surveys.
It's identity-driven, not output-driven. Find what people are protecting before changing what they do.
Every initiative tied to an outcome. Tool choice left to engineers — Claude Code, Codex, Gemini.
AI-adjusted estimation and an outcome-based output dashboard. Bar set at 3–5x on greenfield, 2–3x on legacy.
Ship code daily with the same tools. Credibility beats mandates every time.
Always-on workflows where AI originates ideas, plans, and finished work — and humans approve.
Adoption strategy and change management. Breaking resistance, measuring impact. AI-native workflows, agentic development, developer productivity.
Agent SDK, MCP, orchestration, RAG/embeddings. Structured AI output through deterministic renderers — engineered for >90% build success.
Large distributed teams. Technical vision converted into adoption and delivery. Recruiting, mentoring, succession, executive partnership.
JavaScript, TypeScript, Node.js, Python, Golang. AWS, EKS, Terraform, CI/CD. SOC 2, 99.99% uptime. PostgreSQL, MongoDB, Redis.
The differentiator is the first column. Architecture is table stakes — getting a real organization to actually work this way is the hard part.
Led the AI-native transformation and authored the agentic platform strategy. Scaled engineering to 70+ across 20 countries, then restructured to a leaner, higher-leverage org as AI productivity cut required headcount. Monolith → AWS EKS, SOC 2, 99.99% uptime.
2018 — PresentLed a cross-functional team of 15. Aligned technical strategy with business objectives; company-wide processes for quality, scale, and security.
2016 — 2018Hands-on technical manager from business idea to production. Bridged on-site and remote teams; frontend architecture and code review culture.
2014 — 2016Chose the stack, architected the POC, and built the company's first product from scratch. Remote practices that became company standards.
2011 — 2014Pre- and post-AOL-acquisition. Architected newsGlide, a single-page-application version of HuffPost.
2010 — 2011First engineer on an e-commerce platform, grew to lead a small team of engineers and designers.
2006 — 2010Years in software, 10+ leading at enterprise scale
Peak engineering org, across 20 countries
Uptime, SOC 2 certified, on AWS EKS
Merged PRs per week from a 10-person team
AI-adjusted estimation replaced legacy velocity baselines, and an outcome-based output dashboard keeps the number honest. The 10-person product-engineering team sustains output that previously required a far larger team — 150+ merged PRs a week across a 50+ repository codebase.
"Every initiative tied to an outcome. Tool choice left to the engineer. Leaders ship with the same tools they preach."
I'm renovating my own house — cabinetry, wiring, LED control systems, hidden doors. Our ancestors made physical things every day, and the brain that does abstract work needs that muscle.
Same engineering discipline, different substrate.
Custom cabinets designed and built from scratch — cut lists, soft-close drawers, and a secret compartment behind a pivot-hinged door.
Whole-room addressable lighting on QuinLED/WLED: motion automation, segment presets, power-injection design, 1.4 kW load math. Root-caused an inverted PIR signal at the GPIO level.
Attic closet becoming a mini office: knee-wall cabinets, gable bookshelves, and a full-length mirror door on concealed hinges you'd never find.
Why half-adoption is the most expensive option — and what breaking real resistance inside a real org actually took.
Field notesDaily resistance stories from an organization mid-transformation. The "but" is where the work is.
ArchitectureOne organization of humans and AI agents, staffed by skill — always-on workflows that turn intent into finished work.
Operating model, agentic architecture, and the adoption fight — already fought once at production scale. If your engineering org needs to become AI-native, that's the job I want.
An org that wants to be AI-native, not AI-curious.
Scale — the playbook compounds with headcount.
Leadership that measures outcomes, not activity.
Transformation as the mandate, not a side quest.
Director / Head of AI Engineering · NY metro or remote
Fastest path is email or LinkedIn. Happy to walk through the playbook, the numbers behind it, or how it maps to your organization.