AI Development Framework: Why Evaluation Comes First, Not Later
Why 'we'll optimize later' is the most expensive assumption in AI development. A framework built on evaluation as a first principle, not a phase.
All of my long-form thoughts, collected in chronological order.
Why 'we'll optimize later' is the most expensive assumption in AI development. A framework built on evaluation as a first principle, not a phase.
Hackers ran up $9,000 on my AWS account after a tutorial exposed my credentials. Here's what went wrong and the cost management setup I built before touching AWS again.