# AI Infrastructure from First Principles
This path starts with the core inference loop and builds toward serving concerns.
## Path
1. [[Topics/AI Infrastructure]]
- Start with the topic overview.
2. [[Articles/Inference Core Loop of an Inference Engine]]
- Understand autoregressive generation.
3. [[Articles/Inference Prefill and Decode]]
- Learn why inference has distinct phases.
4. [[Field Notes/KV Cache]]
- Understand the memory-performance tradeoff behind decode.
5. [[Topics/Distributed Systems]]
- Connect inference serving to reliability and distributed systems.
## Related
- [[Topics/Linux & Systems]]
- [[Learning Paths/Zero to Production Engineer]]