Alex Goldhoorn

Projects & Interactive Demos

Personal Projects & Experiments

A collection of code projects and interactive demos exploring optimization, AI, and visualization techniques. These range from production-ready tools to experimental prototypes built for learning and testing new capabilities.

Code Projects

CVRPTW: Vehicle Routing Problem Solver

Personal Learning Project - Python Implementation

  • What it does: Solves the Capacitated Vehicle Routing Problem with Time Windows (CVRPTW), a classic optimization problem in last-mile delivery logistics. Applicable to scenarios like food delivery, e-commerce, and courier services where you need to optimize routes for multiple vehicles with capacity constraints and time windows for pickups and deliveries.
  • Why I built it: Created as a learning exercise to deepen my understanding of constraint programming and Google OR-Tools, building on concepts from my work at Glovo on Live Multibundling and route optimization. This project let me experiment with different solver variants and visualization approaches outside of a production environment.
  • Tech Stack: Python 3.9-3.10, Google OR-Tools, NumPy, Pandas, Plotly, Matplotlib, uv/Poetry, Jupyter notebooks.
  • 🔗 View on GitHub

Discrete Optimization Exercises

Coursera Course Exercises - Python Implementation

  • What it contains: Solutions to the Coursera Discrete Optimization course exercises, covering foundational algorithms for constraint satisfaction, local search, and optimization techniques. Includes implementations for knapsack problems, graph coloring, traveling salesman, vehicle routing, and facility location.
  • Why I did it: Completed as preparation for advanced optimization work in delivery logistics. These exercises provide hands-on experience with optimization algorithms that underpin real-world applications in routing, scheduling, and resource allocation.
  • Tech Stack: Python, various optimization techniques (greedy, dynamic programming, local search, constraint programming).
  • 🔗 View on GitHub

Hide-and-Seek Simulator & Solvers

PhD Research Code - C++ Implementation

  • What it does: Complete simulation and solver framework for robots playing hide-and-seek with humans in urban environments. Implements multiple planning algorithms (MOMDP, POMCP, Particle Filters, Kalman Filters) and includes both console-based and ROS-integrated versions for real robot deployment.
  • Research Context: Core codebase from my PhD thesis (2011-2017) at IRI, CSIC-UPC Barcelona. Used to validate theoretical contributions on planning under uncertainty and was deployed on real robots for experimental validation in multi-floor buildings.
  • Tech Stack: C++11, Qt5, MySQL, OpenCV, Eigen, ROS, APPL 0.95 (POMDP solver), custom Monte Carlo tree search implementation.
  • 🔗 View on GitHub

Research Demos

Interactive visualizations of my academic research work (PhD, Master's, Bachelor's).

Hide and Seek Belief Simulation

PhD Research - Probabilistic Robot Search

  • What it does: Interactive simulation demonstrating belief-based algorithms for robot search in hide-and-seek scenarios. Compare greedy frontier exploration with POMCP (Bayesian grid) and Particle Filter methods. Design custom maps, control the seeker robot manually, and watch how different algorithms handle uncertainty, moving targets, and unknown environments.
  • Tech: Vanilla JavaScript, Canvas API, Real-time Bayesian inference, Particle filtering, Pathfinding (BFS).
  • 📄 Related Paper (2013) | 📚 PhD Thesis

ALV Homing Simulation

Master's Research - Bio-Inspired Robot Navigation

  • What it does: Interactive visualization of the Average Landmark Vector (ALV) homing method inspired by desert ant navigation. Click to add landmarks, set home and robot positions, and watch the robot autonomously navigate home using only the average direction to visible landmarks. Test robustness with different scenarios and compare unit vector vs. distance-weighted approaches.
  • Tech: Vanilla JavaScript, Canvas API, Real-time vector mathematics, MathJax for equations.
  • 📖 About the Method

GNG-SLAM Simulator

Bachelor's Research - Neural Network-Based Mapping

  • What it does: Watch a robot build a map of its environment using Growing Neural Gas (GNG), a self-organizing neural network. The simulation shows how the network dynamically adapts its structure to represent obstacles, automatically allocating more nodes to complex regions and learning topology through lidar data.
  • Tech: Vanilla JavaScript, Canvas API, Real-time neural network visualization.
  • Research: Based on my Bachelor's thesis (2006) on SLAM using self-organizing neural networks.
  • 📖 About GNG-SLAM

Web Tools & Experiments

Personal projects exploring AI, visualization, and rapid prototyping.

Delivery Platform Simulator

Real-Time Logistics Network Simulation

  • Interactive 60 FPS simulation of an on-demand delivery platform modeling order demand, fleet capacity, and urban geography. Compare Greedy vs Hungarian assignment algorithms, visualize courier movement trails, and explore how network effects cause non-linear system behavior during peak demand.
  • Features: Real-time demand modeling with peak hours, fleet efficiency metrics, assignment algorithm performance comparison (O(N) greedy vs O(N³) Hungarian), backlog visualization, and adjustable city size, fleet size, and simulation speed.
  • Tech: React, Canvas API, Discrete Event Simulation, Tailwind CSS.
  • 📖 How it Works | 📄 Related Article (Glovo)

Research Chat (RAG)

AI Chat with My Publications

  • Chat interface powered by Retrieval-Augmented Generation (RAG) that answers questions about my PhD thesis and publications. Uses embeddings to find relevant content and provides cited answers.
  • Tech: Next.js 14, Supabase (PostgreSQL + pgvector), Google Gemini 2.5 Flash.
  • 💬 Try the Chat | 📖 How it Works

Career Knowledge Graph

Interactive Skill Timeline

  • Force-directed graph visualization showing my career evolution from 1999 to 2025 with 100+ skills and technologies.
  • Tech: React, SVG, Custom Physics Simulation.
  • 📖 About | 📄 Data (JSON)

DS Cheat Sheet

Personal Reference Library

  • Personal cheat sheet for Data Science workflows covering optimization algorithms, LLM agent patterns, Python snippets, and more. Features AI-assisted card generation via Gemini API.
  • Tech: React, Google Gemini API.
  • 📄 Data (JSON)

Route Optimizer

TSP Algorithm Playground

  • Interactive visualization of Traveling Salesman Problem (TSP) algorithms. Compare Nearest Neighbor, 2-Opt, Simulated Annealing, Genetic Algorithm, and more.
  • Tech: React, Canvas API.
  • 📖 Algorithm Docs

CVRP Visualizer

Vehicle Routing on Real Maps

  • Interactive visualization of CVRP on a map of Barcelona. Compare algorithms optimizing delivery routes with capacity constraints.
  • Tech: React, Leaflet.js.
  • 📖 About CVRP

Financial Simulator

Interactive Portfolio Modeling

  • Monte Carlo simulations to visualize probability distributions of future portfolio values. Stress-test contribution strategies and asset allocations.
  • Tech: React, Chart.js.

VitalDecision

AI-Assisted Health Triage

  • Symptom tracking tool designed to improve doctor-patient communication. Uses AI to structure health data and provide preliminary guidance on symptom severity.
  • Tech: AI-assisted decision support.

FiraFinder

Catalonia Events & Activities Finder

  • Discover family-friendly events, fairs, markets, and activities across Catalonia. Search by date, region, and category with smart seasonal filters for Christmas markets, Easter events, and local celebrations. Features multilingual support (English, Català, Español, Nederlands) and favorites tracking.
  • Tech: React, Generalitat de Catalunya Open Data API, Tailwind CSS, localStorage.
  • Shortcut: goldhoorn.net/fira

Contact: alex (at) goldhoorn.net