Agentic Systems & AI Agents | Full Stack | Java, Spring, Angular, React, Node.js | Cloud & DevOps
I build AI-native software systems where agents participate in the engineering process and become first-class components of the product itself. Using AI agents, automation, tools and models to turn requirements into working software — while keeping deep traditional engineering knowledge to supervise, integrate and correct code when needed. Architecture, security, validation and observability are decisions I own, not delegate.
Como trabalhar com agentes de IA no dia a dia sem abrir mão da engenharia: intenção humana, agentes, planejamento, ferramentas (MCP), testes, evals, verificação humana e produção — um fluxo completo, da ideia ao deploy.
Do Paint no Windows 98 ao desenvolvimento de software no mundo real. A história de como comecei na programação, os erros no caminho e o motivo por trás deste blog.
AI-native restaurant management platform for restaurants, bars and delivery operations. Comanda eletrônica and PDV, digital menu, order/delivery control, inventory and purchasing, integrated financials, reporting dashboards, role-based access, multi-unit and offline operation. Multi-tenant architecture built with Spring Boot 3 (Java 21), PostgreSQL, Redis real-time events and Angular 21 frontends. Planned evolution: specialized AI agents (Management, Inventory, Financial Insights, Purchasing, Customer Support and Analytics) as first-class components of the system.
Real-time kitchen display system for pizza stores, integrating with the Saipos PDV/ERP via REST API and Firebase Realtime Database. Station-level cooking locks for multi-device teams, voice/print automation, shift lifecycle and a custom delivery routing engine that collects human decisions as training data for a future ML layer.
Multi-store (SaaS) real-time panel that announces ready orders to couriers at the point of exit. Angular 21 + Tailwind + daisyUI frontend, Fastify + Prisma + PostgreSQL + Socket.IO backend with JWT shared across HTTP and WebSocket. Docker Compose stack with Watchtower and Cloudflare tunnel, CI/CD via GitHub Actions.
Complete library management system with JWT auth, book catalog, loans, reservations, admin dashboard, gamification (points, levels, badges, rankings) and a social layer (posts, groups, book of the month). Built with Angular 17 + Spring Boot 3 + PostgreSQL 16 and a daisyUI design system.
Complete pizzeria management platform: orders, deliveries, couriers, employee control, payroll adjustments, vouchers and REST API integrations. Layered architecture (presentation, REST, service, repository, model) built with Jakarta EE 10, JSF + PrimeFaces, Hibernate and MariaDB, containerized with Docker and CI/CD.
Event discovery and promotion wall. Full stack application combining Spring Boot for the API, Angular for the frontend and PostgreSQL for persistence.
Fully responsive ecommerce website for PC components, built with Angular, featuring product catalog, cart and a polished, modern UI.
Minimal and fast URL shortener with click counting and redirect handling.
Mass SMS dispatcher with a simple, intuitive graphical interface for fast, organized promotional campaigns. Built with Python and SQLite, with a local GUI workflow.
Port of 3D Pinball for Windows – Space Cadet to the Sony PSP, re-enabling the game on portable hardware.
JPrinter is a Java application that monitors a specified folder for the creation of text files and automatically sends them for printing using a selected printer. This project is built with JavaFX and leverages Java's WatchService for monitoring file changes and Java Print Service for printing.
The JPix utility is a Java class developed to generate payloads according to the PIX standard (Brazilian Instant Payments System), based on the specifications of the Central Bank of Brazil.
This Python application is designed for conducting Bluetooth Denial-of-Service (DoS) attacks. It allows you to scan for nearby Bluetooth devices and launch a DoS attack against a selected target device.
This script uses the Selenium webdriver to extract phone numbers of members from a WhatsApp group. It opens a WhatsApp Web instance in your default browser, asks you to log in, and then takes a group name as input to extract the phone numbers of all the members in that group.
