Curated learning shelf

LLM Engineering, End to End

Trace the hidden systems behind a dependable AI feature.

Follow four production fault lines—serving memory, retrieved evidence, evaluation blind spots, and tool authority—then learn which measurement or boundary makes each one manageable.

LLM EngineeringThe Memory Bill Behind Every TokenWhy a longer conversation consumes serving memory even after the prompt is processed

Open book

Infographic 5 min

The Memory Bill Behind Every Token

Why a longer conversation consumes serving memory even after the prompt is processed

BBob5w ago

LLM EngineeringRetrieval Has Two Chances to FailA fluent answer can hide whether the evidence was missing or merely ignored

Open book

Infographic 5 min

Retrieval Has Two Chances to Fail

A fluent answer can hide whether the evidence was missing or merely ignored

BBob5w ago

LLM EngineeringOne Score Can Hide the Broken ProductTurn a model benchmark into a map of user promises and failure slices

Open book

Infographic 5 min

One Score Can Hide the Broken Product

Turn a model benchmark into a map of user promises and failure slices

BBob5w ago

LLM EngineeringThe Prompt Is Not a Security BoundaryWhy instructions inside retrieved content must never inherit an agent's authority

Open book

Infographic 5 min

The Prompt Is Not a Security Boundary

Why instructions inside retrieved content must never inherit an agent's authority

BBob5w ago