I build software.
I train AI.

AI water-leak detection

A sensor node that detects leaks in the water network to prevent sinkholes, developed by a team of five for the AI-SED 2026 challenge. My part is the machine learning: time series, light models for microcontrollers and training on the Yuca supercomputer at ACARUS (University of Sonora).

• Python
• Time series
• TinyML
• SLURM · Yuca
• LoRaWAN

Personal AI assistant for an executive

A personal AI assistant for an executive, installed on his computer. It centralises his notes, agenda and to-dos in an Obsidian knowledge base that Claude organises and queries, connected to his sports watch, market prices and LinkedIn. Version 1.0 delivered, with a Windows installer and training.

• Obsidian
• Claude Code
• Python
• GitHub

A team workspace with its AIs connected through MCP

A workspace built from scratch, at zero cost, for a research team: kanban board, documents, chat, proposals with voting and news. Each member’s AI assistant connects through MCP and works on the same data as its person; cards are never deleted, they get archived. In daily use by the team and its advisers.

• React
• TypeScript
• Supabase
• Cloudflare Workers
• MCP

Attendance with QR and official reports

A web system for a public health institution that records nursing students’ attendance at their hospital placements: QR scanning or manual entry, monthly and annual reports in the institution’s official format, PDF badges and role-based access with row-level security. In production.

• Next.js
• TypeScript
• Supabase
• PostgreSQL
• QR

Pricing, inventory, sales and a bazaar store

An installable web app on the iPhone of a jewelry business owner: prices from the real cost, inventory, sales with a scanner and a sale note as an image, a loyalty card and a bazaar store where customers build their cart from their own phone. In production.

• React
• TypeScript
• Tailwind CSS
• Supabase
• PWA

Explainer and product videos from an idea

A system to produce explainer and product videos from an idea: script, synthetic voice, real 3D scenes and kinetic typography, with automatic rendering and a review of the final file. Every frame comes from code, so it can be repeated and fixed.

• HyperFrames
• three.js
• GSAP
• Claude Code
• ffmpeg

AI-powered medication management app

Android app (APK) for tracking medications and dose reminders, with an AI health assistant that answers medication questions using open models such as Gemma and Llama via OpenRouter.

• React Native
• Expo
• TypeScript
• Supabase
• OpenRouter

On-demand home nursing app

Uber-style APK that connects patients with nurses who sign up with their professional license number, for home visits with real-time booking, user authentication and service management.

• React Native
• Expo
• TypeScript
• Supabase
• Real-time

Business management system

Inventory, sales, supplier orders and finances for a real snack business in Hermosillo — synced across the owner’s devices with Supabase.

• Next.js
• TypeScript
• Supabase
• PostgreSQL

Finances, subscriptions, notes and projects in one place

My own command center, the same on the computer and the phone and fully usable offline: finances, subscriptions that warn before they charge, budgets, billing, notes and the real activity of my Claude Code projects. In development: usage, notes and projects come next.

• Next.js
• TypeScript
• Tailwind CSS
• IndexedDB
• Offline sync

How my most recent projects were built

My current method to build real projects with AI agents: a spec before building, multiple-choice questions before each change, agents with explicit model and effort, a reviewer other than the builder and a decision log per project.

• Claude Code
• Agents
• Skills
• Obsidian
• Python