Gabriel Cavedal

Gabriel Cavedal

AI Solutions Architect · Full-Stack & Cloud

Python · Django / DRF · AWS · Cloudflare · Agentic AI

Profile

Infrastructure and software architect; multi-company technical lead for 20+ years. I design systems end to end, working, and coaching teams, in spec-driven development with fleets of AI agents under BDD/TDD discipline. Django/DRF expert. Open to remote consulting: architecture reviews, AI adoption scouting, quick wins.

Experience

Technical Lead — IT & Infrastructure

2016 — present

Grupo ALVS · Mendoza, AR

Lead the group’s IT team across all business units. Designed and operate its AWS cloud platform, which runs the production helpdesk, payment workflows, institutional sites and IoT telemetry over a multi-site VPN backbone. Drive AI adoption group-wide: multi-agent orchestration with MCP servers, automated workflows and hands-on staff training.

Head of Organization & Systems

2005 — 2016

Mendoza Central Entretenimiento (Casino de Mendoza)

Designed the casino’s networks and floorplan end to end and built the RHEL server clusters that ran the entire operation: gaming systems, BI dashboards and machine-learning analytics.

Slot Systems Technician

2002 — 2005

Casino Buenos Aires

Gaming hardware, networks and systems support.

Selected Work

Casino de Mendoza

RHEL clusters · Oracle VM · 700 slot machines

A decade of self-built Linux RHEL clusters behind 700 networked slot machines, with zero downtime.

KCBD — lab & IoT

Raspberry Pi · Django / DRF · Angular

Co-built the cultivation lab’s sensor network; its telemetry streams in production on AWS.

Welp — helpdesk & payments

Django 5 · HTMX · App Runner

Ticketing and purchase-to-payment workflows; in production at Grupo ALVS.

Financial Gateway — group treasury workspace (in progress)

SharePoint Excel · PostgreSQL · Django / DRF · Astro / Svelte · AWS Fargate · Cognito

Collaborative Grupo ALVS build. Finance numbers live in distributed SharePoint workbooks; leadership needs one operational view to decide and act. The product reads those Excel sources and leaves each workbook intact, lands them in PostgreSQL, and serves treasury logic, alerts, and auditable actions through a Spanish interface. Inference classifies operator intent toward a permission-filtered action menu (Amazon Bedrock). Outcome: a precise multi-company treasury surface at fg.grupoalvs.com, with harness and automations keeping product intent across changes.

kodexarg.com — public plaza RAG

Cloudflare Workers · Vectorize · Bun · Astro/Svelte · multi-locale corpus (ES/EN/ZH)

Built and operate a production RAG on the public plaza: visitors evaluating Gabriel for hire get grounded answers from a versioned CV/services corpus—not a demo chatbot. Retrieval over Vectorize; hire-oriented voice (person/service as subject, kodexArg as seal). Rejected: generic “assistant about me” without eval harness. Quality: regression on plaza/hire prose + bias checks after corpus/prompt changes. Outcome: a live, multilingual hire surface tied to cv.kodexarg.com.

Agentic AI tooling

Python · MCP · Antigravity SDK

Public MCP servers, autonomous issue-triage agents and multi-agent orchestration.

Eurotrip 2026

Angular 21 · Cloudflare (Pages · D1 · R2)

Personal travel companion running fully on Cloudflare.

Sociedad Rural del Oeste Argentino

Astro SSR · Svelte 5 · AWS

Spec-driven full-stack build; in production.

Coveris — healthcare SaaS

Angular 21 · Django / DRF · AWS

SaaS built with a multi-tier AI-agent pipeline; MVP on AWS.