Douw Marx's CV

Summary

AI safety researcher working on agentic evaluations, risk estimation, and alignment. Contributing to the European Commission Technical Assistance for AI Safety, Redwood Research, and METR via Equistamp. Recently completed a PhD at KU Leuven, Belgium, applying machine learning and signal processing to fault detection in rotating machines.

Education

University of Pretoria, Honours and Master's in Mechanical Engineering

M.Eng.

Jan 2019 – Dec 2020

KU Leuven, Fault detection in rotating machinery

PhD

Sept 2019 – Oct 2025

Marie Skłodowska-Curie PhD fellow at KU Leuven.

Experience

Equistamp, Research Engineer

Leuven, Belgium

Aug 2025 – present

10 months

KU Leuven, Marie Skłodowska-Curie PhD Fellow

Leuven, Belgium

Sept 2021 – Sept 2025

4 years 1 month

Research on fault detection of rotating machines using unsupervised learning and differentiable signal processing methods.

Wolfram MathCore, SystemModeler Intern

Linköping, Sweden

May 2021 – July 2021

3 months

Develop Virtual labs using Wolfram System Modeler and Mathematica.

XRAM Technologies, Data Analyst and Mechanical Engineer

South Africa

Jan 2021 – Sept 2021

9 months

Build data-driven measurement models for electrode length prediction in electric-arc furnaces.

Wolfram Summer School, Participant

USA

June 2020 – July 2020

1 month

Towards universal robotics

Publications

Talk: Does your LLM care about the same things you do? — 75th Data Science Leuven Meetup

Feb 2026

www.meetup.com/data-science-leuven/events/311853177

Constitutional Sensitivities of Preference Models

Jan 2025

douwmarx.github.io/constitutional_sensitivities_of_preference_models

Patent for a novel parallel kinematic planar mechanism: WO2020208551A1

Feb 2022

worldwide.espacenet.com/patent/search/family/070293014/publication/WO2020208551A1?q=drive%20arrangement%20marx

Scientific articles

Jan 2025

scholar.google.com/citations?user=wSgyJ74AAAAJ&hl=en

Certificates

AI Alignment Course: Bluedot technical alignment course with project: Constitutional sensitivities of preference models.