Daily Reading List – June 16, 2025 (#569)

I’ve been at Google just over five years, but my other job (“dad”) is going on eighteen. The pay is better, hours longer, and purpose stronger. Each Father’s Day is a reminder to prioritize the right job.

[blog] Model Once, Represent Everywhere: UDA (Unified Data Architecture) at Netflix. Having issues with core data concepts being defined differently across systems and teams? So does Netflix. Here’s how they attempted to solve it.

[blog] The Illusion of Vibe Coding: There Are No Shortcuts to Mastery. Lots of these types of articles lately. Nothing inherently untrue here. At the same time, we’re definitely moving up a plane into more of product engineering and architecture. That seems inevitable.

[article] Agent Architect: Orchestrating Intelligent Autonomy. Completed related to the point above. The type of work we’re doing will change.

[blog] Gemini Code Assist adds Gemini 2.5, personalization and context management. This AI coding assistant is on a weekly release cadence now, and keeps getting more useful.

[article] Introduction to vLLM: A High-Performance LLM Serving Engine. This engine doesn’t get the glory of some alternatives, but seems to have mindshare among some very smart folks.

[blog] Mitigating prompt injection attacks with a layered defense strategy. Good post that lays out an in-depth approach to protecting the prompt path in your AI systems.

[blog] Vibe Deployment with AI Studio, Cloud Run, and Jules. I’m not an IDE doomer (yet), but it’s becoming fairly clear that the act of app building and deploying is escaping from the IDE. Good post from Karl.

[article] What comes after Stack Overflow? Matt asks a provocative question. Some answers provided here, including the idea that your input into the LLMs acts as a type of Q&A data source.

[blog] Why Go is a good fit for agents. Java has a headstart on being the second-best option (after Python) for agent builders given the variety of agent framework available. But don’t sleep on Go.

[blog] The lethal trifecta for AI agents: private data, untrusted content, and external communication. Does your agent have these three traits? Be careful, as you’re opened up to a hacker that can cause real mischief.

Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below: