NTN 520 -   Escalar IA Duele
No Tiene Nombre

NTN 520 - Escalar IA Duele

May 19, 2026
23:14
-(0)

About this episode

Servir un LLM no es como escalar una web app normal. Hablamos de tokens, GPUs, KV cache, latencia, costos y por qué “agregar más pods” ya no salva a nadie. Fuentes: • https://www.dheeth.blog/llm-serving-is-not-normal-web-serving/ • https://arxiv.org/abs/2309.06180 • https://kubernetes.io/docs/concepts/workloads/autoscaling/horizontal-pod-autoscale/

Comments

Rate this episode

Episode Details

Show Type
full
Audio Format
audio/mpeg

We use analytics

We use Umami, a privacy-friendly analytics tool, to understand how visitors use this site. No personal data is collected and no data leaves our EU servers. You can accept or decline below. Privacy Policy