About

Over the past four years I've built and led Catalyst, a universal frontend rendering framework that now powers five different business verticals inside 1mg. It started as a way to unify a fragmented codebase. It ended up becoming the platform every team ships on.

The problems I find interesting tend to live at the boundary between JavaScript runtimes and the web platform. Why does this page's TTFB spike under load? Where is memory accumulating in the Node.js rendering process? Why does Webpack's dev server crawl on a cold start? I've chased all of them through profilers, heap snapshots, and flame graphs — the answers were never where I expected.

A few things I'm particularly proud of: streaming SSR with Partial Prerendering inside Catalyst, which cut TTFB by ~35% on SEO-critical routes. A CSS caching rearchitecture — from per-route to shared-file — that reduced runtime memory by 50–70% and eliminated 90–98% of CSS I/O on every render. Deep Node.js heap work that brought pod memory from ~740 MB down to ~440 MB in production, 3.9 GB saved fleet-wide. And a Webpack-to-Vite migration where the solution was writing compatibility layers so teams could adopt it without touching their own code.

Outside the platform work I care about developer experience — Grafana dashboards, OpenTelemetry tracing, and making the platform legible to the engineers building on it. I've mentored six engineers on rendering performance and system design, and I try to write about the things I learn digging into V8 internals. The blog on this site is where those notes end up.

I started at 1mg as a consultant in late 2020 and have been here since — from shipping the Health Records MVP in six weeks to eventually leading the frontend platform team. I studied Computer Science at JSS Academy of Technical Education and graduated in 2020.

If you want to talk about rendering pipelines, runtime performance, or anything in the space between frontend engineering and infrastructure, I'm always up for it.