Skip to content

Technical publication · by William Oliveira

Monday Brief

AI engineering in practice: reliable agents, local models, and efficient systems — from cloud to device.

Issues are planned for Mondays, as time, material and my depression allow.

Subscribe via RSS New issues in your feed reader.

Latest issue

Issue 2

Agent reliability beyond the average

This week: agents that fail on repeat, Pydantic AI security fixes, and SGLang's new storage default. Plus runtime snapshots, reusable context, and audio models running on a phone's CPU.

  • Reliability, evaluation, and observability
  • Security and privacy
  • AI engineering and agent architecture
  • Local AI
  • Edge AI
  • Mobile AI
Read this issue

Previous issues

  1. Issue 1

    Special launch issue: tools, evals, and local inference

    MCP security advisories, agent evaluation pitfalls, and runtime changes in vLLM, Ollama, and Apple's on-device stack. Plus a proposed MLX memory experiment.

    • Security and privacy
    • Reliability, evaluation, and observability
    • AI engineering and agent architecture
    • Local AI
    • Mobile AI