ML and AI systems

This trunk starts at the systems boundary around models: how inference is served, how context is assembled, how agents select and execute actions, and how the whole loop stays observable and answerable to reality.

Start with Agent loops and Inference serving. Use Frontier inference economics when translating large token counters into fresh-prefill load, decode load, API-equivalent spend, and self-host capacity.

The region connects strongly to authority, trust boundaries, retries, latency, and measurement.