Skip to content
ansezz.

▸ Blog series

Serving AI in Production

GPU-aware routing, scaling triggers that actually track LLM load, and the rate limits that stop one agent loop from burning your margin.

← All series
  1. 01

    GPU-aware load balancing for AI inference

    Round-robin breaks when LLM requests span 50 to 50,000 tokens. GPU-aware load balancing with prefill/decode disaggregation and the metrics that cut P99.

    Architecture · 8 min read
  2. 02

    Smart auto-scaling for modern AI apps

    CPU auto-scaling fails GPU workloads. Why queue depth, KV-cache pressure, and TTFT beat CPU as triggers, plus the KEDA patterns to scale AI in time.

    Architecture · 6 min read
  3. 03

    Rate limiting: protecting your AI wallet

    One runaway agent loop can mean a $5,000 LLM bill. Why request-per-second limits lie, and how hierarchical token-bucket limits protect your margins.

    Architecture · 7 min read