AI Letters #13 · Memory Footprint

Memory at Each Stage

Select a stage to see how each framework's RSS changes from import through indexing to active use. Click a framework card to highlight it in the chart.

Import

After Indexing 200 docs

After 50 Queries

Memory composition at this stage

Python baseline (~50 MB)

Framework overhead

Embedding model (~90 MB)

Vector index + data

RSS comparison — all stages