I design and ship AI systems that survive production — multi-agent architectures, advanced RAG pipelines, and on-premise LLMOps/MLOps platforms built on infrastructure I control end to end.
I'm an AI Engineer specialized in designing and building artificial intelligence systems. My focus is on solving complex business problems through secure data pipelines, automated Machine Learning systems, and scalable multi-agent architectures with advanced RAG systems.
My journey started in Full Stack Web Development, evolved into DevOps and Data Science, and now I specialize in Agent AI Ops, LLMOps, and MLOps. I believe the biggest challenge in AI today isn't creating new models — it's bringing them to production securely and profitably.
I design automated pipelines, establish guardrails and sandboxes, and prioritize data sovereignty and GPU infrastructure execution to build reliable, scalable systems for real-world environments — including the self-hosted data center this very site is served from.
“From code to data, and from data to AI”
Production-grade systems, not proofs of concept.
Server-Driven UI platform that translates natural language into fully interactive React components in real time. A 5-node LangGraph pipeline with dynamic MCP tool discovery, hybrid inference (Gemma 2b local + Gemini Flash cloud), and GraphRAG over Neo4j + Qdrant + PostgreSQL. End-to-end latency under 2.5s. Zero hallucinated UI via Pydantic contracts.
End-to-end patient journey orchestration platform — a 24-container intelligence layer deployed over existing legacy systems without disruption. Multi-agent LangGraph pipeline, event bus on Redis Streams, ML anomaly detection (Isolation Forest + XGBoost), and production observability from day one.
LLMOps automation system driven by Genetic Algorithms — autonomously evaluates, mutates, and optimizes LLM inference configurations without human intervention. Reproducible pipelines via Apache Airflow + LangGraph. Transforms manual parameter tuning into a production workflow.
End-to-end supply chain intelligence pipeline on Azure — predictive resampling models coupled to linear programming algorithms, projecting 15–30% reduction in overproduction. Automated BI via Power Automate + live Power BI dashboards. Team lead.
Production home data center built as four independent Docker planes under strict IaC — Cloudflare Tunnel + Traefik v3 edge, CrowdSec + DNS filtering security, centralized multi-host observability (Prometheus · Loki · Grafana), and an always-on platform layer with Wake-on-LAN GPU orchestration. Zero open ports. This very site is served from it.
Local-first research and operations platform for multi-agent systems. Designed for practitioners who need complete observability and control — not a cloud sandbox. Covers agent design, integration pipelines, GPU inference, and structured evaluation workflows.
20-container self-hosted AI lab running RTX GPU-accelerated local inference with Ollama, vector search (Qdrant), ComfyUI, and full LLMOps observability (Prometheus + Grafana + cAdvisor + NVIDIA Exporter). Built under strict IaC and DevOps principles. Zero cloud dependency.
Architected a dual-component enterprise AI system based on Clean Architecture: a self-hosted Multi-Agent System for CRM operations and a local system for confidential institutional archive management. Built a 100% on-premise RAG pipeline using Qdrant and NVIDIA GPU-accelerated local inference (Ollama), guaranteeing absolute data sovereignty.
Containerized IT support platform built for UASLP — real-time ticketing system with RBAC, Laravel + React/TypeScript frontend. Deployed with Docker + Kubernetes, CI/CD pipelines, and real-time monitoring. 99.9% uptime. 40% reduction in incident detection time. 25% faster request resolution.
Awards, launches, and systems that made it to the real world.
Shipped Nahual.AI — a Server-Driven UI platform translating natural language into live React components through a 5-node LangGraph pipeline.
Built CareFlow 360°, a 24-container patient-journey orchestration layer deployed over legacy hospital systems without disruption.
Four Docker planes under IaC — edge, security, observability, platform — behind a zero-open-ports Cloudflare Tunnel. This site runs on it.
NeuroEvolytics: genetic algorithms autonomously evaluating and optimizing LLM inference configurations via Airflow + LangGraph.
Led Linealytics, a supply-chain intelligence pipeline on Azure projecting 15–30% reduction in overproduction.
Architected a dual-component, 100% on-premise enterprise AI system with absolute data sovereignty — multi-agent CRM ops + confidential archive RAG.
Containerized IT support platform with RBAC and CI/CD — 99.9% uptime, 40% faster incident detection, 25% faster resolution.
Tools I reach for in production — not a skill checklist.
Open to full-time, contract, or consulting engagements — remote or on-site, LATAM or international.
Get in touch