Projects

Things I've built and experiments I'm working on.

Midia in the House

Data pipeline that ingests pedestrian flow across São Paulo and turns it into ad campaign decisions. Rust producer → Kafka → BigQuery, with an AI agent (LangGraph + RAG) that reads briefs and ranks media spots. Built with Eletromidia, Brazil's biggest DOOH company.

Rust Python Kafka BigQuery GKE LangGraph

Autonomous Robot Dog

Gave a robot dog a voice and a brain. It listens, thinks, retrieves context via RAG, decides what to do, and talks back. Full speech-to-text and text-to-speech pipeline.

Python RAG STT/TTS Google ADK

blackjack-microkernel-os

An operating system. From scratch. Microkernel architecture in C and Assembly. Has a Blackjack game running on it because why not.

C Assembly OS

holocron

Self-hosted personal dashboard for tracking habits, finances, workouts, diet, and daily life. Full-stack TypeScript + Go + PostgreSQL. Basically my second brain.

TypeScript Go Python PostgreSQL

skills-mgr

One CLI to manage AI coding skills across 19 tools — Claude Code, Cursor, Copilot, Gemini, you name it. Fuzzy search, symlinks, zero drama.

Shell CLI AI Tools

SOD — Optical Detection

Fly a drone over a building, get back a map of every crack. AI-powered structural monitoring with segmentation models, a Next.js dashboard, and a Kotlin drone app. Built with IPT.

Python Next.js Kotlin Supabase Computer Vision

adalov3

Chrome extension that makes my college's platform (Adalove) actually usable. Students use it daily. Published on the Chrome Web Store.

JavaScript CSS Chrome Extension

Teia — LEI Website

Built the website for Inteli's Entrepreneurship League — the public landing page plus all the internal tools the league runs on. TypeScript frontend, Go backend.

TypeScript Go Full-stack

EstiMatch

Real-time match predictions for Brazilian soccer using ML. Built in partnership with IBM — predictive models that actually work on live games.

Python ML Predictive Analytics IBM

The Six Club

Matchmaking algorithm for networking dinners. You tell it who's coming, it figures out who sits next to who. 3,000+ people matched before the startup wrapped up.

Python Algorithms Automation