Hi! I'm Ryan.

Biomedical Engineering @ Waterloo.

I explore the art of peeling back the hood by building systems from end-to-end.

Currently, I'm conducting AI research into Deep Causal Transformers. On the side, I'm doing machine learning and data engineering for a pre-seed startup. I've also built Ryan-GPT, a full-scale language model training pipeline with tools to simulate distributed training on a single GPU.

Lately, I've been thinking about:

  • how representation learning differs between language and vision models
  • how modern machine learning systems are constructed
  • conviction: taking an idea and seeing it through to completion

This site is a quiet record of what I'm building and thinking about.

Reach out anytime!

Recent thoughts