(Note: the code snippets do not look great in this article. I’m publishing for now because I want to get the content out. I plan to adjust the styling later) In Figure 1 below, you will see a photo of a “mobile” thin-client version of my personalized development environment. Look closely. You will see it is running on an iPad. I have found the iPad to be pretty good for on-the-go development because a) it is light b) has long battery life c) has a great screen d) is cheaper than a typical developer-spec laptop and e) there are no secrets or code are stored locally presenting less of a security headache if the device is lost or stolen.

In this series, I’d like to share my experiences of over 20+ years of architecting large complex enterprise systems, products and services. If you are involved in any kind of software development endeavor in whatever role, hopefully you find something of value in this series. The first question you may immediately ask is, if some of these guidelines are borne from experiences from years ago, surely they are out of date?

In July 2023, I knocked up a simple AI chatbot using the open source Llama model from Meta. I coded it up from scratch using LangChain. It had the ability to build a vector store from your own proprietary or sensitive documents, enabling you to chat “over them”, without risking un-intentional disclosure. It ran reasonably ok, using a quantized version of Llama, but the main selling point was the complete privacy and zero subscription costs.