I'm an engineer at Cohere. I lead our data infrastructure and data quality work – building resilient data pipelines for neural networks and researching the effects that different kinds of training data have on large transformer models. I've recently been interested in critically examine modelling benchmarks to better align with the tasks we hope for them to perform. Here's some recent work.
Previously, I was at Shopify where I worked on Kubernetes ingress and routing infrastructure for a fleet of 50+ clusters (fun fact: it's one of the largest Kubernetes deployments in Google Cloud) and on Sections Everywhere, the biggest update to the Online Store Editor in the history of the company. Here's a talk I did about some of it.