Begin at Starbucks

Table of Contents

LLM_Engineer - This article is part of a series.

Part 1: This Article

My reading starts at Starbucks… 🗽

Chapter-1
#

This book, LLM Engineer’s Handbook, begins with the why building an LLM Twin. And this is the ultimate goal of this book.

Then it continues by introducing MVP (minimum viable product) which is what features will be delivered. This is a powerful strategy, and the key is, sticking to the V in MVP.

Last, it introduces the solution – ML pipelines for ML systems. It is based on the FTI (feature, training, inference) pattern.

This is the most valuable concept I learn in chapter-1.

Imagine this, with FTI, I can start explore into Google Cloud architecture document.

The FTI pipelines act as logical layers to interact with each other through the feature store and model registries. It is a tool used to clarify how to design ML systems.

This shows how FTI pattern has been applied to design the architecture of a ML system (LLM Twin).

Excellent job! From why to what, and from what to how. And I repeat this chapter twice.

Chapter-2
#

Chapter-2 begins with preparing the Python ecosystem and project installation. This makes me wonder should I switch from uv to poetry for the sake of convenient.