Full-stack platform for creating, managing, and growing brands in a freelance environment. Built across two phases: first as a marketing web app with a project management framework, then refactored with a graph architecture and migrated from React to Next.js. Backed by MongoDB and a GraphQL API.

Website for a Saarbrücken-based research lab studying mobility culture. Designed and built as a freelance project with a focus on clean information architecture.

Rental listing website for a portfolio of apartments in Saarbrücken. Built with Next.js for fast static rendering and straightforward property browsing.

Platform for Synsen — a creative consulting collective and artist community exploring the intersection of synthetic sense and technology. Serves as the digital home for the group's projects, artists, and events.

Online presence and product catalogue for a company that imports and distributes Bolivian specialty foods. Built with Next.js, designed to communicate the brand's cultural identity.

Website for a Spanish-speaking theater group based in Saarbrücken. Jorge has served as director of the group's 2024, 2025, and 2026 seasons, working with multicultural actors on production, organisation, and programming.

Coherence-enhancing diffusion (CED) is a PDE-based anisotropic smoothing technique which selectively smooths along flow structures while preserving edges. Studied as part of Saarland University's image processing coursework.

Implementation on pytorch of the Coherence enhancement diffusion algorithm optimized for GPUs.

Project for the seminar "Deep Generative Diffusion Models " - Saarland University 2023.It uses generative difussion

Echo State Networks are a special type of neural network that only need to train a small part of their internal structure, while still maintaining accuracy. This makes them faster to train and less demanding on memory compared to traditional neural networks. This paper explores whether this advantage can be applied to image classification (teaching a computer to recognize what's in a picture).

This paper summarizes the key ideas behind a technique that reconstructs 3D surfaces of objects using multiple 2D images, by separately handling the shape of an object and its visual appearance (like color and texture).

This paper uses CLIP to tackle two specific challenges applied to movie posters: identifying the genre(s) of a movie based on its poster (e.g. action, comedy, horror), and automatically generating a catchy tagline for it.

Implementation of the mandelbrot set fractal optimzed using PyTorch for GPUs.This is an abstraction for the iterative method for constructing a fractal. As such, it allows to define custom iterative function. For this case the function used corresponds to the mandelbrot set, but it can extended to other sets like Julia

Co-authored analysis of the energetic potential of landfill sites across different regions of Bolivia, assessing the viability of biogas as a local energy source. Published in UPB Revista Investigación y Desarrollo.

Co-authored research demonstrating the effectiveness of reservoir computing — a recurrent neural network approach — on multivariate product demand forecasting. Presented at the POMS 2018 conference and published by Springer.

Co-authored paper exploring dictionary learning as a technique for decomposing and reconstructing temporal signals. Shows that learned dictionary atoms can act as complex, data-driven filters — offering an interpretable alternative to black-box preprocessing for time series. Published in UPB Revista Investigación y Desarrollo.

Gallery of Julia set fractals generated with custom shaders over the years. Each image explores how small changes in the complex parameter c produce radically different fractal geometries.

Zoom animation into the Burning Ship fractal — a variant of the Mandelbrot set with a distinctive flame-like structure. Rendered with a Python pipeline; the same fractal drives the animated background of this portfolio.

Zoom-out animation of the Sierpinski triangle, revealing the self-similar structure that repeats identically at every scale.

Parametric animation of spirograph variations — curves traced by a point on a circle rolling inside another circle, producing intricate geometric patterns from a single simple rule.

Step-by-step construction of the Sierpinski triangle through an iterative space-filling curve, showing the emergence of fractal structure from a simple recursive rule.

Coherence-enhancing diffusion applied to a Monet painting and animated to apreciate the smoothing effect

Collection of handmade paintings — part of an ongoing creative practice alongside the technical work.