Engineering Manager and Software Architect at Moneris in Toronto. I have eight years of experience across fintech, cleantech and utilities, including five years leading software teams. I grew CarbonCure's software team from two people to twelve, then hired an eight-person team from scratch at Moneris. Most of my work is in TypeScript and Node.js on cloud backends, and I run a self-hosted LLM server at home on two AMD MI50 cards.
- Go-Typescript-React-Boilerplate and Electron-Typescript-React-Boilerplate: starting points I use for new projects.
- Pulumi-DigitalOcean-ContainerWebsite: infrastructure as code for hosting a container on DigitalOcean behind a load balancer and a domain.
- drizzle-neon: Neon Postgres tables managed with Drizzle in a Node.js TypeScript project.
- ninfer-gfx906 (closed): a port of the NInfer inference engine to AMD Instinct MI50 cards. It works, but llama.cpp is faster on this hardware, and the README explains why. My server runs mx-llama.cpp, and I send that project MI50 bug reports and test data.



