The Stack · nine layers, one pair of hands
The Stack
An AI product is built at every one of these layers at once. This is where I work — each layer cites the shipped work that proves it.
Bottom to top: from the machine to the business. One tick per shipped work.
Layer 01 · Cloud · deploys · CI/CD · observability
Infrastructure & Operations
I set up the pipelines and environments my products run on, and I'm the one watching them after they ship.
Evidence · 5 shipped
GitHub Actions → ECR → ECS deploys, owner-gated production behind a manual workflow, Terraform for the whole production estate, 31 CloudWatch alarms and synthetic canaries.
Deploys to EKS through GitHub Actions, with Terraform, CDK, and Helm, and Grafana dashboards, SLOs, and alerts.
GitHub Actions run the validation gauntlet, evals, and a benchmark on every change, a nightly benchmark on a schedule, and publish Storybook to GitHub Pages.
Packaged it as a signed macOS app and installer a colleague runs from one file — nothing hosted, customer data stays on the machine.
CI runs lint, typecheck, tests, contract validation, import-boundary checks, and migrations; Terraform modules define ECS, RDS, ElastiCache, SQS, and observability.
Trace a work through the stack
The index, in full
09
Enterprise & Business Architecture
I define systems and transformations at enterprise scale — and the products inside them are built by the same hands.
08
Product Strategy & Roadmap
I decide what gets built first and why, and I own the outcome rather than the backlog.
07
Research & Insight
I start from how people behave, not how a roadmap assumes they will.
06
Experience & Interface Design
I design the surface people actually touch, and I care about how it reads as much as how it works.
05
Design Systems
I build systems of tokens and components that both people and models can compose without breaking.
04
Product Engineering
I write the front end myself, from the first prototype to the version people use.
03
Agentic Runtime & Architecture
I design the machinery that decides what an agent may do, and what happens after it does.
02
Data & Model Layer
I wire models into products and decide what context they see, then check the output against something other than my own impression.
01
Infrastructure & Operations
I set up the pipelines and environments my products run on, and I'm the one watching them after they ship.