[ AI & PLATFORM ENGINEER — LLM APPS · AGENTS · DISTRIBUTED SYSTEMS ]
Salvador
F. Criado
I build production AI end to end — agents, RAG and real-time voice, plus the GPU inference underneath. Eight years owning distributed systems on AWS and Azure that stay up.
8+ yrssystems in production
<2.0svoice loop, end to end
100k+users on my platforms
3clouds: aws · azure · on-prem
01 / What I do
Agents & LLM apps
RAG, tool calling, voice loops under 2 s — grounded in your data.
Serving & evaluation
vLLM, quantisation, Langfuse tracing — quality you can measure.
Cloud backbone
Deep AWS, expert Terraform, Kubernetes — the platform under the model.
02 / Selected work
full history on linkedin →-
On-prem voice agents for an enterprise contact centre
Agentic flows, RAG on Qdrant and a streaming voice loop under 2 s — on the client's own GPU, data never leaves the network.
2025—26 -
AWS → Azure migration of a 100k-user e-learning platform
Multi-tenant platform lifted onto AKS: Terraform, Azure DevOps pipelines, and a cutover with the service staying up.
2026
03 / Writing
Field notes
The half of your GenAI system nobody put on call
Nobody's RAG system breaks — it quietly stops being right. The data jobs nobody wrote down, the three mechanisms that make them safe, and what to put on a dashboard.
Temperature Zero Is Not Deterministic
Your RAG doesn't have a retrieval problem. It has a ranking problem.
New notes land in the RSS feed. No newsletter, no signup.
04 / About
Hands-on engineer in Granada, Spain — remote worldwide. I started close to the metal, on PCBs and firmware, and worked up through distributed systems to platforms carrying a hundred thousand users. Applied AI is the newest layer on that stack, not a replacement for it: I still reach for Terraform and a profiler before I reach for a bigger model. AWS DevOps Pro · Agentic AI (DeepLearning.AI) · CSM.