Good end to the week. I appreciate having co-workers that I can just call up and strategize with. Or complain too. Or celebrate with. All at the same time.
[youtube-video] The Agent Factory – Episode 1: Agents, their frameworks and when to use them. This new podcast/video series has promise. It features some colleagues who did a good job looking at the overall landscape, and zeroing in on some Google innovations.
[article] Stop forcing AI tools on your engineers. Very good advice here. Definite give teams time to explore, give space, and know what matters. Also, don’t wait forever.
[blog] Beyond GROUP BY: Introducing advanced aggregation functions in BigQuery. This seems like a good deal for data folks. The post calls out some big performance and efficiency benefits of these new aggregation functions.
[blog] Our new approach to enterprise CI/CD: Free tier, source available, and guaranteed savings. There’s still interesting work happening in the CI/CD space. The Semaphore folks are making some moves.
[blog] User Count: One. Bespoke Software with Gemini CLI. You can just build software for yourself. Billy wanted help focusing, so built a Chrome extension for an audience of one.
[article] AI coding tools may not speed up every developer, study shows. I’m not just going to read (and share) posts that say everything is sunshine. We’re still all figuring out where these AI tools add value, and when other fundamentals need to be in place for them to be useful.
[article] Ollama or vLLM? How to choose the right LLM serving tool for your use case. “When to use what” is such an important question in so many areas nowadays. Here’s help with LLM serving tools.
[blog] Google Brings the Lustre Parallel File System to Its Cloud. Teams are trying to squeeze out every bit of performance on their ML jobs, and this should help them do that, with less management.
[blog] SQL reimagined: How pipe syntax is powering real-world use cases. You should give this a whirl. In reading the example queries, it does seem to have some real benefits over standard SQL.
[blog] Graph foundation models for relational data. This seems like some creative thinking around ML algorithms and training data.
[blog] How to use GenAI as an Executive Assistant. Lak wrote a series of seven posts (so far) about different roles that you can apply generative AI to. Check out his other posts about engineering, research, analyst, and more.
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