AI-Native Design¶
Pages in This Section¶
| Page | Covers |
|---|---|
| Model Serving | Batching (static vs. continuous), KV cache, GPU autoscaling/cold starts, quantization, cost model |
Out of scope for this site, by design: context windows, scheduling, routing/gateways, RAG, hybrid search, reranking, embeddings, vector DBs, tools/agents, evals, guardrails, hallucination mitigation. That material lives in a dedicated AI engineering resource: sanketn26.github.io/AIEngineering.
This section stays narrow on purpose — Model Serving covers the systems/infra side of running models (batching, KV cache, autoscaling, cost) because that's squarely a distributed-systems problem, which is this site's lane. Application-layer AI engineering is a different discipline with its own resource.
Next: Model Serving →