Alex Goldhoorn

Data Scientist • Robotics Researcher

About Me

Alex Goldhoorn

Senior Data Scientist at Meight with 12+ years of industry experience and a PhD in Robotics from UPC Barcelona. My work spans logistics optimization, discrete event simulation, LLM agents, machine learning, and robotics.

Previously at Glovo, I led the architecture of a discrete event simulator serving 120,000+ couriers across 23 countries and built optimization pipelines that reduced delivery times network-wide.

My PhD research focused on planning under uncertainty with noisy sensor data—skills applicable to autonomous systems and complex optimization problems. I'm particularly interested in opportunities in space, transportation, and high-tech industries where these approaches create real-world impact.

Beyond work, I'm a proud husband and father who enjoys running (road and trail), riding motorcycles, and good food from my Dutch and Catalan backgrounds.

More about me →

Highlighted Work

Freight Optimization & LLM Agents

Meight (2024-Present)

  • LLM Agents & Evaluation: Designed the evaluation framework for AI Agents, measuring faithfulness, tool usage, and cost for both structured and unstructured outputs. This ensures reliability for agents automating admin tasks for freight managers.
  • Tour Optimization: Developing algorithms to optimize freight planning and reduce empty running. This involves integrating processed real-time telematics and EU driving-time regulations to boost Revenue Per Kilometer (RPK) for carriers.

Delivery Simulator & Route Optimization

Glovo (2019-2024)

  • Delivery Simulator Leadership: Led the architecture and refactor of the discrete event simulator. This tool was the optimization and testing ground for the dispatch and matching algorithms, allowing the team to tune hyperparameters and validate logic for a network spanning 23 countries and 120k+ monthly couriers.
  • Matching Optimization: Built the automated optimization pipeline for the matching cost function, enabling rapid experimentation to balance delivery speed, courier earnings, and operational efficiency.
  • Route Optimization (VRP): Led the VRP implementation for bundling, enabling the system to dynamically batch orders and optimize routes in real-time.

Read technical writeup → Delivery Simulator Demo →

PhD & Research: Robotics & Uncertainty

IRI, CSIC-UPC & University of Groningen (2006-2017)

  • PhD Thesis: Searching and Tracking of Humans in Urban Environments. Developed algorithms (MOMDP, POMCP) for robots to find people, treating the problem as a "Hide-and-Seek" game under uncertainty.
  • MSc Thesis: Solving Ambiguity in Global Localization. Focused on resolving robot position ambiguity in symmetrical environments using probabilistic sensor models.
  • BSc Research: Early work on SLAM (Simultaneous Localization and Mapping) using Growing Neural Gas (GNG) networks to map unknown environments.

View thesis → Interactive demos → 💬 Chat with my research →

Explore interactive simulations of ALV homing, GNG-SLAM mapping, and vehicle routing algorithms. Or try the AI-powered chat to ask questions about my publications using RAG (Retrieval-Augmented Generation).

View Full CV → Download CV (PDF)

Highlighted Projects

Code projects, research demos, and personal tools exploring optimization, AI, and interactive visualizations. View all projects →

Professional & Research Projects

CVRPTW Solver

Personal project exploring vehicle routing optimization for last-mile delivery. Built to experiment with Google OR-Tools and constraint programming techniques.

Research Chat (RAG)

AI chat interface for my publications using Retrieval-Augmented Generation. Ask questions about my robotics research and get cited answers.

DS Cheat Sheet

Personal reference library for data science workflows with AI-assisted card generation. Python snippets, algorithms, and patterns.

Hide & Seek Belief

Interactive demo from my PhD research. Compare probabilistic search algorithms (POMCP, particle filters) for robot navigation.

ALV Homing

Bio-inspired robot navigation from my Master's research. Interactive simulation of desert ant navigation algorithms.

Personal Tools

🏃 RaceFinder

Find and plan your next running adventure. Browse 500+ road and trail races worldwide with advanced filtering, favorites tracking, and race comparison tools.

Career Knowledge Graph

Interactive force-directed graph showing 25 years of skills and technologies. Explore 100+ nodes with timeline filtering and search.

FiraFinder

Discover family-friendly events, fairs, and markets across Catalonia. Multilingual support and smart seasonal filters for local celebrations.

Featured Articles

Technical writings on AI evaluation, simulation, and optimization. View all articles →

When LLMs Meet Structured Data: The Evaluation Challenge

January 2026

Building a two-track evaluation framework for LLM agents at Meight. When extracting structured shipping data, we learned that evaluation requires both strict metrics (for production readiness) and LLM-as-a-judge (for semantic correctness). Neither alone is sufficient.

Read Article →

System 1 vs System 2: Testing LLMs with Riddles

December 2025

An experimental evaluation of how 8 models (6 cloud, 2 local) perform on logic puzzles, revealing the gap between pattern matching and first-principles reasoning. Includes complete raw model responses and an interactive challenge.

🎯 Try Interactive Challenge → Read Full Analysis → 📝 Raw Outputs →

How to Simulate a Global Delivery Platform

February 2021 • Medium

Deep dive into building a large-scale discrete event simulation system for Glovo's global delivery network, covering architecture decisions, performance optimization, and real-world validation.

Read on Medium →

Get in Touch

Email: alex (at) goldhoorn.net

Interests

LLMs & AI Agents Transportation & Logistics Optimization Machine Learning Data Science Discrete Event Simulation Vehicle Routing Robotics Running Travel Investing