Search posts, tags, users, and pages
Priya Raman
Exploring AI tooling, LLM applications and what breaks in production.
We spent two weeks rewriting prompts. Then we changed the response format, and the failure rate dropped by half. That is not a story about prompt engineering being useless. It is a story about measuri
Markdown supported · ⌘↵ to post
Daniel Okafor
Building developer tools and writing about the friction nobody demos.
Building the eval set first is the part everyone skips under deadline pressure.
Daniel Okafor
Building developer tools and writing about the friction nobody demos.
Building the eval set first is the part everyone skips under deadline pressure.