I'm Gonçalo, a Computer Engineering student from Portugal 🇵🇹.
I'm still early in my journey, and honestly, that's the fun part — everything here is an experiment. I build small things, break them, read the error messages (eventually), and come back knowing a bit more. Main quest: finish the degree. Side quest: build something that survives production.
I've been using AI since ChatGPT first went public in 2022, so by now it's baked into how I study and build. Daily drivers: Claude and Claude Code. I've also worked with Codex on GPT models up to SOL 5.6, run Qwen and Hermes locally, and experimented with OpenClaw. Beyond the models themselves, I build custom skills for Claude and wire AI into bigger workflows with tools like n8n and Obsidian. The tools change every month — the skill I'm actually training is learning fast with them, without letting them do the thinking for me.
And I don't just chat with models — I build with them. A local multi-agent pipeline that produced Reels/TikTok/Shorts end-to-end (script, images, voice, final cut with animations) and auto-published to YouTube through its API. Jarvis, a personal desktop agent that opened apps and ran actions on my computer. And right now, an AI trading bot — my current main experiment.
these builds live off-GitHub for now — repos coming as I clean them up 🧹
┌─ NOW LOADING ────────────────────────────────────────┐
│ ▸ Django ........... deeper into models, forms, auth │
│ ▸ Next.js + TS ..... components, routing, deploys │
│ ▸ LLM tooling ...... agents, APIs, local models │
│ ▸ Git .............. branching without fear │
└──────────────────────────────────────────────────────┘
[ USED IN PROJECTS ] Python · Django · JavaScript · TypeScript
React/Next.js · HTML · CSS · SQLite · Git
[ AI TOOLBOX ] Claude · Claude Code · Codex/GPT · Qwen
Hermes · OpenClaw · Claude skills · n8n · Obsidian
[ LEARNING NOW ] Better Django patterns · TypeScript · REST APIs
[ WANT TO EXPLORE ] LLM APIs & agents · Docker · C
|
|
|
|
|
|
Bonus round: LLMRockPaperScissors — a fork where an LLM plays rock-paper-scissors. I poked at it to see how Java talks to a language model. The LLM still loses to random chance sometimes, which is oddly comforting.
- Survive another semester of Computer Engineering
- Build something real with AI agents (see
$ ls ai/builds/above) - Get the trading bot through a full week of paper trading
- Open-source one of my AI builds
- Write a proper README for every repo (the irony is not lost on me)
week 38 — git push origin hope
↑ this quote rotates weekly via GitHub Actions, so the profile never fully sleeps.
🎮 PRESS START — hidden level
ACHIEVEMENT UNLOCKED: You clicked a collapsed section on a stranger's profile.
That's exactly the kind of curiosity I respect.
Fun fact: my first "it works!" moment was followed, 30 seconds later,
by my first "why does it not work anymore?!" moment. Still chasing that ratio.
The terminal is always open:
- 💼 LinkedIn: gonçalo-oliveira
- 📧 Email: goncaloro@gmail.com
- 🐙 Right here on GitHub