← Paths
03
Scale and Infrastructure
What separates a system that works from one that works at scale, under load, at a viable cost.
6 lessons · ~80 hours
0 / 6
- 01
Context engineering
Retrieval, ranking, compression, position — as an engineering surface.
- 02
Loop engineering
Termination, retries, self-correction. Most agent failures are loop failures.
- 03
Multi-agent systems
Supervisor, hierarchical, blackboard — and what each coordination pattern costs.
- 04
Distributed agents
Message passing, shared state, and partial failure across machines.
- 05
Inference optimization
Quantization, batching, KV caching. Where the latency actually goes.
- 06
AI infrastructure
Serving, scaling, and fine-tuning economics — your unit costs.
