We’ve got a day off at Google tomorrow and I plan on doing some light reading and meeting up with friends. Still expect a reading list, though!
[article] LlamaIndex review: Easy context-augmented LLM applications. This is a good writeup about one particular LLM orchestration framework, while also helpfully referencing other options too.
[blog] Human I/O: Detecting situational impairments with large language models. All of us face “situational impairments” when we’re in a noisy restaurant, have soap in our eyes while taking a shower, or driving with a phone on the seat next to us. This is very cool research.
[blog] How to create software quality. “High quality” is a loaded term. Quality means different things based on the context, and your context. Will’s thoughtful post might change your thinking about it.
[docs] Context caching. This is a BIG deal. We shared that this was coming, but now you can cache context into the Gemini model, which means you’re not passing in 1m tokens each time.
[blog] The Complete Guide to Tables in Google Sheets. Here’s a fairly comprehensive look at structured tables in your cloud spreadsheets.
[blog] 6 deployment archetypes for your reliability, cost, operational, and latency needs. It’s not hard to choose a cloud architecture that’s ill-suited to your requirements. Use guides like this to pick the right model for your requirements.
[blog] Become a Better Java Developer: 19 Tips for Staying Ahead in 2024. Java is a mature, stable language with a mature, stable ecosystem. But it’s also easy to “fall behind” on important updates to both. This looks like a good list.
[blog] BigQuery Federated Queries with Oracle. The ability to reach into databases from the data warehouse is powerful. It’s also much simpler than having to transfer all the data to a central place for analysis.
[blog] BigQuery community UDFs go global to simplify data transformations for everyone. These open source user defined functions can now be accessed from any of our cloud regions. And you can contribute other ones to the repo.
[blog] RAG is simpler and more powerful than you think. Aja does a nice job explaining this LLM pattern and why it matters.
##
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 
One thought