On a whim, I tried asking Vertex AI to generate an image of a “Richard in an office” and was disappointed that similar requests for “Scott” and “Cheng” looked a LOT cooler than “Richard.” We’re going to need to shut down this AI thing until we can get to the bottom of it.
[blog] No GPU? No problem. localllm lets you develop gen AI apps on local CPUs. Here’s a new (open) tool and process for working with LLMs on machines that don’t have GPUs.
[blog] The Importance of Attending Conferences. Online learning doesn’t replace what you get from in-person gatherings. The networking is better, the learning is more focused, and the ideas are more free-flowing. It’s also more expensive. But, companies should value the long-term investment.
[blog] Visualize PaLM-based LLM tokens. The LLM-mania has really sparked some great creativity. My colleague built an app that helps you understand token counts in your prompts.
[article] 4 Secrets Of High-Performing Teams. Give this a good read and internalize these points around team structure, effectiveness, interactions, and leadership.
[blog] Using Vertex AI Gemini REST API (C# and Rust). Every API doesn’t have a corresponding SDK. That’s ok, but it does mean you have to dig deeper to understand the low-level interactions, even including authentication. Mete shows a useful example here.
[article] Can Enterprise DevOps Ever Measure Up? DevOps has been around a while. Why are big companies still having a tough time making progress on key measures? Do we even agree on the measures? This article explores.
[docs] Regional deployment on Compute Engine. As much as I like fully-managed compute platforms, I recognize that most of the world’s software is running on piles of VMs. This new architecture guide looks at deploying an app across zones in a cloud region.
[blog] DotSlash: Simplified executable deployment. This new open source tool from Meta saves them from sticking platform-specific executables next to projects that rely on them.
[blog] Practical Guide to Generating Embeddings in AlloyDB with Google ML Integration. I’m still in the “reading about” phase with embeddings and vector databases. This post is helpful whether you’re like me, or actually starting to implement solutions.
##
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