Skip to content

ruben@portfolio:~$ whoami

Systems engineering student, building toward AI engineering & research.

I’m a self-directed builder from Armenia, Colombia. I learn by shipping — command-line tools, web apps and AI tooling — turning genuine curiosity into working software.

01 — About

Learning by building.

I’m a second-semester Systems Engineering student. I learn in cycles, driven by genuine curiosity rather than obligation — building practical projects is how concepts actually stick for me.

Right now I work in the Technology & AI department at VAECOS, my family’s dropshipping business, automating real operational processes. My long-term aim is to become an AI Engineer and Researcher — someone who both understands how these systems work and can build genuinely useful things with them.

I think in long, ambitious arcs, and I care about contributing to my community, not just personal success.

02 — Skills

Tools I build with.

Self-taught, project-first. This list grows with every repo.

Languages

Python TypeScript JavaScript Java Bash

Frontend

React Astro Tailwind CSS

Backend & Data

Node.js PocketBase SQLite

AI / ML

Vercel AI SDK Ollama RAG Claude API

Tooling & Infra

Docker Git Linux (Arch) Neovim Vitest

03 — Work

Selected projects.

A short list on purpose — the work I stand behind, not everything I have ever pushed.

// note

Every project here uses AI-assisted development: I own the architecture, the decisions and the testing, and I level up my hands-on coding with each one. Knowing how to direct AI to ship real software is a skill in itself — one I practise deliberately.

2026

A desktop app that records, transcribes and summarises classes and meetings entirely on your own machine. Built on top of the open-source call.md, swapping its paid cloud dependency for local Whisper transcription (sherpa-onnx) and a pluggable LLM backend — Ollama or any OpenAI-compatible provider. Ships a working Windows installer, 257 automated tests and its own landing page.

ElectronTypeScriptReactsherpa-onnxOllama
/02

VAECOS Tracking

2026

The internal logistics and finance platform running in production for our family dropshipping store, which operates in Guatemala. It reconciles shipment tracking across systems, creates shipping labels end to end in a third-party ERP through a browser bot with an AI address validator, keeps the financial ledger, and answers questions about its own data through a conversational assistant with tool-calling. It has been running the daily operation of the store — around 30 orders a day — since 2026.

PythonFlaskSQLitePlaywrightLLM tool-calling
Private repository — internal to the business
See all repositories on GitHub ↗

04 — Contact

Let’s build something.

I’m open to collaboration, open-source contributions and learning opportunities. The fastest way to reach me is email.