Upskilling Your Tech Stack

Explain how tokenization, context windows, inference, and probabilistic generation affect a local document assistant.

Distinguish semantic retrieval, retrieval-augmented generation, tool use, and autonomous agent behavior using document-search examples.

Create a threat model covering sensitive source files, extracted text, embeddings, logs, temporary files, and model outputs.

Diagram a local-first architecture connecting a React client, Python API, document pipeline, search index, agent runtime, and local model.

Run a quantized language model through Ollama on an Apple Silicon Mac and measure its memory use, latency, and output quality.

Expose a minimal local-model request through a FastAPI endpoint and consume its streamed response from the existing React project.