-
Before the Breakthrough: Why Research and Engineering Need Different Cultures
Research is not slow engineering. Why companies that want breakthroughs need to cherish two cultures, not collapse them into one.
-
The Bayesian Story Behind Prior-Fitted Networks
PFNs are often described as Bayesian predictors, but their training objective and inference mechanism suggest a more nuanced interpretation.
-
Why Uncertainty in Machine Learning Is Conceptually Broken
A critique of why modern ML uncertainty estimates lack clear semantics, reliable evaluation, and meaningful use cases.
-
Code Sharing at Scale in Python Monorepos with a Single Version Policy
How structuring internal modules as installable packages in a monorepo enables controlled code reuse across many containers — while keeping images lean and dependency management centralized.
-
Why your next visual quality inspection system will be training-free
Why visual inspection should rely on approaches that do not require images of defective products for training.