Notes on applied ML — synthetic data, evaluation, and the parts of deployment that papers leave out.
Versioning data, experiments, models, and environment on a small team that resists process.