ernesto mendoza blanco.
Writing

Technical writing

Agentic AI

  • When to read AI-generated code

    · 4 min read

    A practical rule for deciding when AI-generated code deserves a line-by-line review, a test-first review, or a quick scan.

  • Evals are the new load tests

    · 6 min read

    Reproducible evals are to agent quality what load tests were to latency: a practical gate on the failures that only appear at the tail.

Leadership

  • Leading engineers who ship with LLMs

    · 4 min read

    Shipping with LLMs is a leadership problem before it's a tooling one: teams need accountability, review culture, and repeatable gates.

Observability

  • The anatomy of a good observation

    · 9 min read

    A framework from Norwu's observability playbook for turning measured evidence into an observation a team can understand and use.

  • What p99 hides and how to see it

    · 4 min read

    Averages lie, percentiles compress, and p99 still hides shape. Histograms make the tail visible enough to debug.