Some thought-provoking opinions in a few of the items in the list below. I’m realizing how quickly my own opinions evolve as a saturate myself in different ideas.
[blog] Why Writing Matters (to the Technical Professional). From what I can tell, good communicators succeed more often than bad ones. Keep working on the craft of writing!
[blog] One Toolbox, Many Sources: Unifying Your Data Landscape with MCP Toolbox. We’re going to end up with MCP sprawl, yes? Seems inevitable. Solutions like this which let you talk to multiple databases via a single MCP server might be the right pattern.
[article] Top findings from the 2025 Stack Overflow Developer Survey. Good look at a few dimensions of how engineers work.
[article] What New CEOs Should Ask Themselves in Their First 100 Days. This is written for CEOs, but I thought it was useful for someone in any new leadership role.
[blog] The 5 Most Common REST API Design Mistakes (and How to Avoid Them). I’ve probably done all of these. This week alone. But good reminders that we should pay attention to.
[blog] Designing API Error Messages for AI Agents. I’ve never seen someone talk about this. What kinds of error messages should you return to help an agent figure out next steps?
[blog] Inferencing — Serve Llama 4 on A4 (B200 GPUs) using vLLM and GKE. Good instructions on serving an LLM on a fully-managed Kubernetes cluster.
[article] Can Platform Engineering Accelerate AI Adoption? The recent PlatformCon had a lot of discussion about platform engineering, and Jennifer pulls out many key insights.
[blog] Introducing Gemma 3 270M: The compact model for hyper-efficient AI. Here’s a flavor of our open model family that’s great for fine-tuning.
[paper] TextQuests: How Good are LLMs at Text-Based Video Games? Can modern models perform well when playing text-based video games? This research puts that idea to the test.
[article] GPT-5: A Choose Your Own Adventure for Frontend Developers. Fascinating launch of this model. Different experiences are still coming out. Theo revised his early impressions.
[blog] Designing a multi-tenant GKE platform for Yahoo Mail’s migration journey. There are some useful details in this story about Yahoo’s migration to the cloud.
[article] New trend: extreme hours at AI startups. There are usually seasons where you need to dig in and work extra. We’ve all been there. But making it your culture? Red flag.
[blog] How Keeta processes 11 million financial transactions per second with Spanner. Wow, some impressive numbers for a blockchain and payments business.
Want to get this update sent to you every day? Subscribe to my RSS feed or subscribe via email below:
Richard Seroter's Architecture Musings 
Leave a comment