Hi, I’m Keegan.
I’m a Principal Software Engineer. I build and scale TypeScript and Node.js services on Kubernetes, and these days I spend most of my time helping engineering teams adopt AI-native development so they can deliver faster.
I’ve worked across the whole stack — frontend, backend, data, and infrastructure — and led teams as an engineering manager and CTO.
AI-native development,
hardened for teams.
My team’s pipeline starts out like any good CI/CD pipeline: automated checks, validation steps, and deployment gates. What’s changing is how much of the work AI does inside it, and that takes new guardrails.
That’s what I’m building: rules for how AI reviews code and how AI-contributed pull requests are made, and tooling that lets agents inspect the surrounding system and show that new code fits into it correctly.
The goal is an agent that can take a feature from start to finish and prove the result before anything merges: checked for correctness, and for alignment with what the business is actually trying to achieve.
This site is built the same way: one command builds, lints, type-checks, and tests every page in three screen sizes, and produces screenshots an agent can review.
From search monolith
to microservices.
Selected work
Over a couple of years, I led a team in moving a legacy search monolith to microservices built with TypeScript and Node.js on OpenSearch and Redis on ElastiCache. It’s the most business-critical system I’ve worked on: the product it powers brings in most of the company’s subscription revenue, and it serves 1,000–1,500 requests per second at peak.
Moving off a monolith is something many teams start and few finish. This one finished, and the new services carry that traffic today.
Less bespoke code,
less to maintain.
Selected work
With my team, I replaced a legacy, bespoke commenting framework in a free-form application with a general-purpose pattern built on standard React components. The app kept its commenting, and the team stopped maintaining a one-off framework: less code to carry, and a codebase that’s easier to work in.
TypeScript and Node.js on top,
Kubernetes underneath.
Most of my work is TypeScript on Node.js: APIs and services that need to stay fast and dependable under real load. I describe infrastructure as code with Terraform and Helm, store and search data with OpenSearch, Redis on ElastiCache, and DynamoDB, and scale applications in Kubernetes.
I like TypeScript because it balances flexibility and safety, but the right language is the one that makes the team and the business effective. Over the years I’ve shipped production code in Go, Ruby on Rails, C#/.NET, Java, PHP, and more.
I still love making things move.
I started out building interactive work for the web, and I still take a special satisfaction in math-driven animation, generative art, and 3D. The cubes behind this page are one of my experiments, and there are more on the way.
A few technologies I’ve picked up just for the challenge: