Welcome

This is where I will write about the things I am working on, mostly computer vision and deep learning, with a bias toward evaluation: what a metric actually measures, where a model quietly fails, and what it takes to trust a result outside the benchmark it was tuned on. Expect notes from projects in progress rather than finished conclusions.