ENTREPRENEURSHIP

AppLinkers

Web platform built from scratch for university communities. Used by @linkers.madrid and @salseo_usc, with a combined reach of 200,000+ followers. Started as a social matching app with a vectorized compatibility algorithm (NumPy) and intelligent filters (exclusive, reciprocal, bonus), and has grown into a full university toolkit: apartment and room search with compatibility-based matching, permanent housing-compatibility test and per-community event management.

Relational data model with multi-tenancy and reusable question hierarchies with full match traceability. GDPR compliance with versioned consent, anonymization preserving referential integrity and automated rights (access, erasure, art. 22). Infrastructure with three independent environments (production / dev / local) with audit system on its own DB, encrypted automated backups and tooling to preserve critical data when copying between them. Email-based claims with auto-resolution tokens, address obfuscation and transactional templates. PWA with full dark mode, per-environment feature flags, matching-algorithm calibration system with synced sliders, live metrics refreshed every 8h on the dot and cron jobs for scheduled tasks. Housing module from scratch: filtered search with custom dropdowns, owner dashboard with stats, editor with gallery, Google Maps Geocoding, GTFS public-transport data integration with Haversine nearest-stop calculation, OAuth signup for owners and room request flow. Operations managed via custom CLI and TUI wrapper (OpenTUI + SolidJS) bridging Python subprocesses.

DjangoPostgreSQLNumPyCeleryRailwayAPIPWAOAuthGDPR

Homelab

Self-hosted infrastructure on Proxmox VE with multiple VMs and LXC containers. Reverse proxy, monitoring stack (Grafana, Prometheus), Docker applications, and automated backup systems.

ProxmoxDockerLinux

@aeropedrax

Free and open educational initiative for engineering students. YouTube channel with lecture-style videos and shared course notes. Strong engagement and lasting positive impact among students.

YouTubeEducation
OPEN SOURCE

Parallel Processing

Research on GPU parallel computing strategies for matrix and vector operations. Benchmarking and performance tuning of scientific computations across multiple languages. The bachelor's thesis extended this with a novel PDE formulation using global Lagrange interpolation. Built so that any reader can reproduce every benchmark on their own machine with step-by-step instructions.

CUDAGPUHPC