Systems software, AI tooling, and self hosted products.
Brazil-based developer working across Go, Rust, Swift, Python, TypeScript, and C++.
I like software that can be server'ed and run anywhere just with a browser, or native tools that actually solves a real problem:
- Local-first tools that keep data and control on the user's machine
- Self hosted systems
- AI workflows that produce inspectable, reproducible output
- Developer tools for storage, agents, telemetry, and automation
| Project | What it is |
|---|---|
| Tchebets | Experimental pari-mutuel market platform with separate services, accounts, wallets, and devnet plumbing. |
| Econowatt | Adaptive GPU energy, thermal, fan, and power control for Linux hardware. |
| Juycy | JuiceFS manager with non-destructive sync, recovery, and a native macOS shell. |
| LNsources | Native macOS resource monitor and local telemetry recorder. |
| Asseter | Bounded AI workbench for inspectable SVG, raster, and web assets. |
| Learnix | AI study workspace with adaptive quizzes, notes, citations, and saved progress. |
| Dubbline | Local-first video dubbing pipeline for transcription, translation, voice synthesis, and timing. |
| Shakehands | Linux evdev service for filtering keyboard bounce and accidental mouse clicks. |
- Keep the backend modular to work with any frontend
- Prefer local and reversible operations.
- Keep boundaries explicit between experimental and production code.
- Ship working slices, then harden them with tests and real usage.
Some repositories are early engineering snapshots, most are not polished at all to be honest. Their READMEs state their current maturity and intended safety boundaries.
Go Rust Swift Python TypeScript C++ SQLite Postgres Redis Docker Tauri SwiftUI
