I worked for 8 years as a Principal Engineer at Shopify, chasing bottlenecks in every layer of the stack from 100s of RPS to ~1M. I’ve seen every stage of scale: from startup, to powering a significant chunk of the world’s commerce.

Today, I work with multiple companies to scale and evolve their web infrastructure.

I can deliver the most value if your problem fits one of the following:

1. Audit. You bring me your pressing infrastructure woes, I write a detailed, actionable report with actionable solutions that your engineers can run with. 1-3 weeks
2. Prototype. You bring me a gnarly backend/infrastructure/database problem, I iterate from first principles to implement you a prototype your team can run with. Either alone, or with part of your team. ~2 months
3. Launch. You’re launching and worried your infrastructure might crack under the pressure. I do an audit, hop on the ground to firm up the infra for launch, and give you a post-launch infra roadmap. 2-4 weeks
4. Review. You’re about to make a big infrastructure decision, I go through and review it with you or the team in detail. ~1-2 days
5. Workshop. You want to level up your infrastructure team in system design. I do a workshop in-person or online on first-principles systems thinking and napkin math. ~2 days

Make your infrastructure decisions with someone who’s done it before.

Next step is to email simon@webscale.ca and tell me which problems you’re dealing with, and we’ll set up an intro call.

Examples

• 2023
  • Prototype for Replicate: faster cold boots for large ML models
  • Audit for Typefully: focus on Timescale scalability, best-practises, and observability.
• 2022
  • Review for Whimsical
  • Launch for Readwise: Reader launch, especially database and job tuning
  • Prototype for Readwise: Ghostreader, AI-reading assistant
  • Prototype for Datafold: data-diff, fast diff between two databases
  • Review for Rutter: Database migration woes
  • Prototype for Causal: Scaling their spreadsheet engine from millions to billions of cells

I am comfortable at all layers of the stack, from Rails to React to the Kernel. You can read more about what I spend time thinking about on the Napkin Math blog.