Today had a good mix of meetings and creative time. That’s rare for a Thursday. Did you get to create something today?
[article] Delivering Great Developer Experiences with Platform Engineering. This virtual panel interview looks at why engineering teams built shared platforms, what hurdles they faced, how they measured the impact, and what they’d do differently.
[blog] Get to know Vertex AI Model Monitoring. Are your model’s predictions drifting away from what is happening in production right now? This new feature helps you monitor your ML models (running on any serving infrastructure) and flags anomalies you can use to trigger re-training.
[blog] Maintaining large-scale AI capacity at Meta. With this kind of infrastructure, it takes some new thinking to keep it running at scale. Good post from the Meta team.
[blog] How Meta trains large language models at scale. Another one from the Meta engineering team, this time looking at how they’re pushing boundaries on software and infrastructure.
[blog] Understanding Konfig’s Opinionation. If you have a bare bones source control and CI/CD strategy, you might be motivated to pivot hard to something more opinionated. Tyler’s post here explains a very opinionated, but well thought-out, platform setup.
[blog] The Four Kinds of Research-and-Development Teams. Ted wrote up a series of posts that explain each type of R&D team, their flaws, and how to succeed.
[blog] AI in software engineering at Google: Progress and the path ahead. Here are some good details from Google research folks into what we’re learning about AI assistance in software development.
[article] How does access to conversational AI impact software engineering trust and productivity? This looks at some recent research and looks at how engineering managers might apply what’s been uncovered.
[blog] 5 LLM Fine-tuning Techniques Explained Visually. If you don’t know your LORA from your VERA, read this post. But more importantly, get a quick overview of different ways you can fine-tune a model.
[article] Microsoft kills off Copilot GPT Builder after just 3 months. There’s lots of experimentation happening now, even in vendor land, so invest accordingly!
[article] Send Personalized Emails using Gemini API, SendGrid and Node.js. Interesting use case. Create and send more personalized email by using an LLM.
##
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